Claims:WE CLAIM:
1. A method for in silico prediction of one or more chemical pathways for transforming one or more start compounds to one or more target compounds, comprising:
multi-directionally predicting one or more output molecules, through one or more reaction prediction steps, produced by each of given one or more inputs using a knowledgebase,
wherein the one or more inputs comprise one of the one or more start compounds, the one or more target compounds and the predicted one or more output molecules,
wherein the one or more output molecules produced at a previous reaction prediction step become input for next reaction prediction step, and
wherein the one or more output molecules being predicted at each of the reaction prediction step by applying a set of one or more transformation rules, included in the knowledgebase, on the one or more inputs;
collecting all the multi-directionally predicted one or more output molecules into a set of one or more intermediate molecules, after every reaction prediction step;
computing similarity between the multi-directionally predicted one or more output molecules within the set of one or more intermediate molecules to create one or more groups of similar one or more output molecules and identify representative member for each of the one or more groups of similar one or more output molecules,
wherein the computation is performed after each of the reaction prediction steps or after a preset number of the reaction prediction steps;
replacing each of the one or more groups of similar molecules within the set of one or more intermediate molecules with the single representative member, thereby using the representative member of each of the one or more groups of similar molecules and non-similar multi-directionally predicted one or more output molecules as one or more inputs for next reaction prediction step; and
connecting together the one or more start compounds, the one or more target compounds, the multi-directionally predicted one or more output molecules, and sequence of the one or more reaction prediction steps to predict the one or more chemical pathways, thereby reducing redundancy in pathway prediction computation and computed pathway data,
wherein the one or more predicted chemical pathways comprise sequential arrangement of the one or more reaction prediction steps governed by the one or more transformation rules.
2. The method as claimed in claim 1, wherein number of the reaction prediction steps for the one or more start molecules and the one or more target molecules are preset.

3. The method as claimed in claims 1 and 2, the one or more reaction prediction steps are repeatedly performed on the one or more inputs until total number of the reaction prediction steps equals to the preset number of the reaction prediction steps from the one or more start molecules and one or more target molecules.

4. The method as claimed in claim 1, where the knowledgebase comprises list of reactions, one or more chemical moieties present in each of the listed reaction, change taking place in the one or more chemical moieties present in the listed reactions during the reaction, set of one or more transformation rules governing each of the enlisted reaction, and the set of one or more transformation rules represented by a unique list of one or more transformations.

5. The method as claimed in claim 1, further comprising sorting the predicted one or more chemical pathways based on at least one of reaction feasibility, kinetics and abundance of intermediates formed in the chemical pathway.

6. The method as claimed in claim 5, further comprising selecting one or more chemical pathways out of the sorted one or more chemical pathways based on at least one of reaction feasibility, kinetics and abundance of intermediates formed in the chemical pathway.

7. A device for in silico prediction of one or more chemical pathways for transforming one or more start compounds to one or more target compounds, comprising:
a memory; and
one or more processors operatively coupled to the memory, the one or more processors are configured to perform the steps of:
multi-directionally predicting one or more output molecules, through one or more reaction prediction steps, produced by each of given one or more inputs using a knowledgebase,
wherein the one or more inputs comprise one of the one or more start compounds, one of more target compounds and the predicted one or more output molecules,
wherein the one or more output molecules produced at a previous reaction prediction step become input for next reaction prediction step, and
wherein the one or more output molecules being predicted at each of the reaction prediction step by applying a set of one or more transformation rules, included in the knowledgebase, on the one or more inputs;
collecting all the predicted one or more output molecules into a set of one or more intermediate molecules, after every reaction prediction step;
computing similarity between the predicted one or more output molecules within the set of one or more intermediate molecules to create one or more groups of similar one or more output molecules and representative member for each of the one or more groups of similar one or more output molecules,
wherein the computation is performed after each of the reaction prediction steps or after a preset number of the reaction prediction steps;
replacing each of the one or more groups of similar molecules within the set of one or more intermediate molecules with the single representative member, thereby using the representative member of each of the one or more groups of similar molecules and non-similar multi-directionally predicted one or more output molecules as one or more inputs for next reaction prediction step; and
connecting together the one or more start compounds, the one or more target compounds, the multi-directionally predicted one or more output molecules, and sequence of the one or more reaction prediction steps to predict the one or more chemical pathways, thereby reducing redundancy in pathway prediction computation and computed pathway data,
wherein the one or more predicted chemical pathways comprise sequential arrangement of the one or more reaction prediction steps governed by the one or more transformation rules.
8. A method for in silico prediction of one or more chemical pathways for transforming one or more start compounds to one or more target compounds, comprising:
identifying one or more chemical moieties from the one or more start compounds and the one or more target compounds received as input,
wherein the identification of chemical moieties is performed with reference to a knowledgebase for a pair of the one start compound and the one target compound, formed from the input, at a given time;
extracting the identified one or more chemical moieties pertaining to a functional group from each pair of the start and target compound received as input;
constructing a functional chemical moiety vector representing functional groups for the one or more start and target compounds received as input based on the knowledgebase;
computing difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of the input start compound;
computing transformation vector based on the computed difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of input and a rule transformation matrix,
wherein the input is one of a one or more start compounds and one or more intermediates produced from the input one or more start compounds after application of a set of transformation rules enlisted in the transformation matrix;
identifying sequence of transformations from at least one of the one or more start compounds and the one or more intermediates to the one or more target compounds based on the computed transformation vector; and
predicting the one or more chemical pathways based on the sequence of the transformations identified while applying corresponding one or more transformation rules, present in the knowledgebase, unto the one or more start compounds and intermediates to predict the one or more target compounds, wherein
the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules.
9. The method as claimed in claim 7, wherein the knowledgebase comprises list of reactions, one or more chemical moieties present in each of the listed reaction, change taking place in the one or more chemical moieties present in of the listed reactions during the reaction, one or more transformation rules governing each of the enlisted reaction, and the set of one or more transformation rules represented by a unique list of one or more transformations.

10. The method as claimed in claim 7, wherein the rule transformation matrix is derived from the knowledgebase and comprises of rows correspond to the chemical moieties and columns correspond to the transformation rules.

11. The method as claimed in claim 7, wherein the rule transformation matrix is populated based on predefined rules where
negative entry in the matrix denotes that the one or more transformation rules acted on the identified chemical moiety leading to one of deletion and modification of the identified chemical moiety,
positive entry in the matrix denotes that the one or more transformation rules acted lead to the formation of the chemical moiety, and
zero entry in the matrix denotes no effect brought upon the identified chemical moiety by application of the one or more transformation rules.
12. The method as claimed in claim 7, wherein the one or more intermediates produced at a previous chemical reaction prediction step become input for the next chemical reaction prediction step.

13. A device for in silico prediction of one or more chemical pathways for transforming one or more start compounds to one or more target compounds, comprising:
a memory; and
one or more processors operatively coupled to the memory, the one or more processors are configured to perform the steps of:
identifying one or more chemical moieties from the one or more start compounds and the one or more target compounds received as input,
wherein the identification is performed with reference to a knowledgebase for a pair of the one start compound and the one target compound, formed from the input, at a given time;
extracting the identified one or more chemical moieties pertaining to a functional group from each pair of the start and target compound received as input;
constructing a functional chemical moiety vector representing functional groups for the one or more start and target compounds received as input based on the knowledgebase;
computing difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector the input start compound;
computing transformation vector based on the computed difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of input and a rule transformation matrix,
wherein the input is one of a one or more start compounds and one or more intermediates produced from the input one or more start compounds after application of a set of transformation rules enlisted in the transformation matrix;
identifying sequence of transformations from at least one of the one or more start compounds and the one or more intermediates to the one or more target compounds based on the computed transformation vector; and
constructing the one or more chemical pathways based on the sequence of the transformations identified while applying corresponding one or more transformation rules, present in the knowledgebase, unto the one or more start compounds and intermediates to predict the one or more target compounds, wherein
the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules.
14. A method of compactification of one or more chemical pathways for transforming a given one or more start compounds to one or more target compounds, comprising:
acquiring a plurality of chemical pathways for transforming the given one or more start compounds to one or more target compounds, wherein the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules based on a knowledgebase;
identifying at least one of
a plurality of the reactions acting on same reactant and product pair where plurality of the chemical reactions being predicted using different transformation rules, and
a plurality of chemical pathways, formed by same set of the transformation rules, predicted for the given one or more start compounds to one or more target compounds, and
a plurality of chemical pathways, having similar intermediates, predicted for the given one or more start compounds to one or more target compounds;

grouping together at least one of
the identified plurality of the reactions acting on the same reactant and product pair, and
at least one of the one or more chemical pathways formed by the same set of transformation rules and the one or more chemical pathways having similar intermediates; and

compactifying the one or more groups of the one or more chemical pathways for transforming the given one or more start compounds to one or more target compounds.

15. The method as claimed in claim 14, further comprising assessing the compactified groups of one or more chemical pathways based on predefined parameters.

16. The method as claimed in claim 15, wherein the predefined parameters for assessing the compactified groups of one or more chemical pathways comprises at least one of a physio-chemical and one or more statistical properties.

17. The method as claimed in claim 14, further comprising removing cycles of at least one of intermediates and transformations from the acquired plurality of chemical pathways based on the identification of at least one of the plurality of the reactions acting on same reactant and product pair, the one or more of chemical pathways formed by same set of the transformation rules, and the one or more of chemical pathways having similar intermediates.

18. The method as claimed in claim 14, wherein the knowledgebase comprises list of reactions, one or more chemical moieties present in each of the listed reaction, change taking place in the one or more chemical moieties present in of the listed reactions during the reaction, one or more transformation rules governing each of the enlisted reaction, and the set of one or more transformation rules represented by a unique list of one or more transformations.

19. A device for compactification of one or more chemical pathways for transforming a given one or more start compounds to one or more target compounds, comprising:
a memory; and
one or more processors operatively coupled to the memory, the one or more processors are configured to perform the steps of:
acquiring a plurality of chemical pathways for transforming the given one or more start compounds to one or more target compounds, wherein the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules based on a knowledgebase;
identifying at least one of
a plurality of the reactions acting on same reactant and product pair where plurality of the chemical reactions being predicted using different transformation rules, and
a plurality of chemical pathways, formed by same set of the transformation rules, predicted for the given one or more start compounds to one or more target compounds, and
a plurality of chemical pathways, having similar intermediates, predicted for the given one or more start compounds to one or more target compounds;

grouping together at least one of
the identified plurality of the reactions acting on the same reactant and product pair, and
at least one of the plurality of chemical pathways formed by the same set of transformation rules and the plurality of chemical pathways having similar intermediates; and

compactifying the one or more groups of plurality of chemical pathways for transforming the given one or more start compounds to one or more target compounds.
20. The device as claimed in claim 19, wherein the one or more processors are further configured to perform the step of:
assessing the compactified groups of plurality of chemical pathways based on predefined parameters.

Dated this the 26th day of October 2015
Signature

KEERTHI JS
Patent Agent
Agent for the Applicant

, Description:FORM 2

THE PATENTS ACT, 1970
(39 OF 1970)
&
THE PATENTS RULES, 2003
COMPLETE SPECIFICATION
(Section 10 and Rule 13)

Method and device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s)

SAMSUNG R&D INSTITUTE INDIA – BANGALORE PRIVATE LIMITED
# 2870, ORION Building, Bagmane Constellation Business Park,
Outer Ring Road, Doddanakundi Circle,
Marathahalli Post, Bangalore-560 037

An Indian Company

The following Specification particularly describes the invention and the manner in which it is to be performed
FIELD OF THE INVENTION
The present invention relates to knowledge-based expert systems and computational synthetic chemistry, and more particularly relates to in silico prediction of one or more chemical pathways for transforming input of one or more start compounds to one or more target compounds
BACKGROUND OF THE INVENTION
Synthesis or degradation of chemicals through chemical and/or biochemical pathways require assessment of large chemical space. Given complexity and vastness of prediction, identification and validation of a proposed synthetic pathway is pursued through in silico simulations. In silico identification of novel synthetic chemical pathways requires two components. One, a robust collection/library of possible chemical transformations governed by a plurality of reaction rules, and two, an efficient system, termed chemical transformation processor, which uses the reaction rules to transform input target molecules and predict new product molecules.
The in silico simulation process for predicting novel synthetic chemical pathways involves application of the reaction rules from reaction rule library on a target molecule and predicting set of products or precursors (retro-synthesis). To generate multi-step chemical pathways, the process is iterated on all predicted products/precursors to find out the individual reaction steps comprising the chemical pathway. In order to select an experimental tractable synthetic pathway the process is repeated for multiple steps (order) to obtain an appropriate end/start compound. Unfortunately, computational intensiveness increases exponentially with each iteration. Thus the simulations produce a large data set which is difficult to mine and almost impossible to manually inspect and assess.
Wherefore, there is need for an improved method and device for in silico prediction of chemical reactions which could reduce and refine simulation data for pragmatic assessment.
SUMMARY OF THE INVENTION
The present invention provides for methods and devices for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s).
In an embodiment of the present invention, a method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) is disclosed. The method starts with multi-directionally predicting output molecule(s), through reaction prediction step(s), produced by each of given input(s) using a knowledgebase. The input(s) include start compound(s) and/or target compounds and the predicted one or more output molecules. The output molecule(s) produced at a previous reaction prediction step become input for next reaction prediction step. The output molecule(s) being predicted at each of the reaction prediction step by applying a set of transformation rule(s), included in the knowledgebase, on the input(s). This is followed by collecting all the multi-directionally predicted output molecule multi-directionally predicting output molecule(s) into a set of intermediate molecule multi-directionally predicting output molecule(s), after every reaction prediction step. Further to this, computing similarity between the multi-directionally predicted output molecule multi-directionally predicting output molecule(s) within the set of intermediate molecule multi-directionally predicting output molecule(s) to create group multi-directionally predicting output molecule(s) of similar output molecule multi-directionally predicting output molecule(s) and identify representative member for each of the group multi-directionally predicting output molecule(s) of similar output molecule multi-directionally predicting output molecule(s). The computation is performed after each of the reaction prediction steps or after a preset number of the reaction prediction steps. Further to this, replacing each of the group multi-directionally predicting output molecule(s) of similar molecules within the set of intermediate molecule multi-directionally predicting output molecule(s) with the single representative member, thereby using the representative member of each of the group multi-directionally predicting output molecule(s) of similar molecules and non-similar multi-directionally predicted output molecule multi-directionally predicting output molecule(s) as multi-directionally predicting output molecule(s) inputs for next reaction prediction step. Finally, connecting together the start compound multi-directionally predicting output molecule(s), the target compound multi-directionally predicting output molecule(s), the multi-directionally predicted output molecule multi-directionally predicting output molecule(s), and sequence of the reaction prediction step multi-directionally predicting output molecule(s) to predict the chemical pathway multi-directionally predicting output molecule(s), thereby reducing redundancy in pathway prediction computation and computed pathway data. The predicted chemical pathway multi-directionally predicting output molecule(s) comprise sequential arrangement of the reaction prediction step multi-directionally predicting output molecule(s) governed by the transformation rule(s).
In another embodiment of the present invention, the method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) further includes sorting the predicted one or more chemical pathways based on at least one of reaction feasibility, kinetics and abundance of intermediates formed in the chemical pathway.

In yet another embodiment of the present invention ,the method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) futher includes selecting chemical pathway(s) out of the sorted chemical pathway(s) based on at least one of reaction feasibility, kinetics and abundance of intermediates formed in the chemical pathway.
In another embodiment of the present invention , a device for in silico prediction of one or more chemical pathways for transforming one or more start compounds to one or more target compounds is disclosed. The device includes a memory, and processor(s) operatively coupled to the memory. The processors are configured to perform the steps including: (a) multi-directionally predicting output molecule(s), through reaction prediction step(s), produced by each of given input(s) using a knowledgebase. The input(s) include start compound(s) and/or target compounds and the predicted one or more output molecules. The output molecule(s) produced at a previous reaction prediction step become input for next reaction prediction step. The output molecule(s) being predicted at each of the reaction prediction step by applying a set of transformation rule(s), included in the knowledgebase, on the input(s); (b) collecting all the multi-directionally predicted output molecule multi-directionally predicting output molecule(s) into a set of intermediate molecule multi-directionally predicting output molecule(s), after every reaction prediction step; (c) computing similarity between the multi-directionally predicted output molecule multi-directionally predicting output molecule(s) within the set of intermediate molecule multi-directionally predicting output molecule(s) to create group multi-directionally predicting output molecule(s) of similar output molecule multi-directionally predicting output molecule(s) and identify representative member for each of the group multi-directionally predicting output molecule(s) of similar output molecule multi-directionally predicting output molecule(s). The computation is performed after each of the reaction prediction steps or after a preset number of the reaction prediction steps; (d) replacing each of the group multi-directionally predicting output molecule(s) of similar molecules within the set of intermediate molecule multi-directionally predicting output molecule(s) with the single representative member, thereby using the representative member of each of the group multi-directionally predicting output molecule(s) of similar molecules and non-similar multi-directionally predicted output molecule multi-directionally predicting output molecule(s) as multi-directionally predicting output molecule(s) inputs for next reaction prediction step; and (e) connecting together the start compound multi-directionally predicting output molecule(s), the target compound multi-directionally predicting output molecule(s), the multi-directionally predicted output molecule multi-directionally predicting output molecule(s), and sequence of the reaction prediction step multi-directionally predicting output molecule(s) to predict the chemical pathway multi-directionally predicting output molecule(s), thereby reducing redundancy in pathway prediction computation and computed pathway data. The predicted chemical pathway multi-directionally predicting output molecule(s) comprise sequential arrangement of the reaction prediction step multi-directionally predicting output molecule(s) governed by the transformation rule(s).
In yet another embodiment of the present invention, a method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) is disclosed. The method steps starts with identifying chemical moiety(ies) from the start compound(s) and the target compound(s) received as input. The identification of chemical moieties is performed with reference to a knowledgebase for a pair of the one start compound and the one target compound, formed from the input, at a given time. Further to this, extracting the identified chemical moiety(ies) pertaining to a functional group from each pair of the start and target compound received as input. Further to this, constructing a functional chemical moiety vector representing functional groups for the start and target compound(s) received as input based on the knowledgebase. Next step includes computing difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of the input start compound. Further to this, computing transformation vector based on the computed difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of input and a rule transformation matrix. The input is start compound(s) or intermediate(s) produced from the input start compound(s) after application of a set of transformation rules enlisted in the transformation matrix. Next to that, identifying sequence of transformations from at least one of the start compound(s) and the intermediate(s) to the target compound(s) based on the computed transformation vector. Finally, predicting the chemical pathway(s) based on the sequence of the transformations identified while applying corresponding transformation rule(s), present in the knowledgebase, unto the start compound(s) and intermediate(s) to predict the target compound(s). The chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by transformation rule(s).

In a further embodiment of the present invention, another device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) is disclosed. The device includes a memory, and processor(s) operatively coupled to the memory. The processors are configured to perform the steps including: (a) identifying chemical moiety(ies) from the start compound(s) and the target compound(s) received as input. The identification of chemical moieties is performed with reference to a knowledgebase for a pair of the one start compound and the one target compound, formed from the input, at a given time; (b) extracting the identified chemical moiety(ies) pertaining to a functional group from each pair of the start and target compound received as input; (c) constructing a functional chemical moiety vector representing functional groups for the start and target compound(s) received as input based on the knowledgebase; (d) computing difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of the input start compound; (e) computing transformation vector based on the computed difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of input and a rule transformation matrix. The input is start compound(s) or intermediate(s) produced from the input start compound(s) after application of a set of transformation rules enlisted in the transformation matrix; (e) identifying sequence of transformations from at least one of the start compound(s) and the intermediate(s) to the target compound(s) based on the computed transformation vector; and (f) predicting the chemical pathway(s) based on the sequence of the transformations identified while applying corresponding transformation rule(s), present in the knowledgebase, unto the start compound(s) and intermediate(s) to predict the target compound(s). The chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by transformation rule(s).

In yet another embodiment of the present invention, a method of compactification of chemical pathway(s) for transforming a given start compound(s) to target compound(s) is disclosed. The method includes the steps of - (a) acquiring a plurality of chemical pathways for transforming the given start compound(s) to target compound(s), wherein the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules based on a knowledgebase; (b) identifying at least one of a plurality of the reactions acting on same reactant and product pair where plurality of the chemical reactions being predicted using different transformation rules, and a plurality of chemical pathways, formed by same set of the transformation rules, predicted for the given one or more start compounds to one or more target compounds, and a plurality of chemical pathways, having similar intermediates, predicted for the given one or more start compounds to one or more target compounds; (c) grouping together at least one of the identified plurality of the reactions acting on the same reactant and product pair, and at least one of the one or more chemical pathways formed by the same set of transformation rules and the one or more chemical pathways having similar intermediates; and (d) compactifying the one or more groups of the one or more chemical pathways for transforming the given one or more start compounds to one or more target compounds.

In another embodiment of the method of compactification of chemical pathway(s), the method further includes assessing the compactified groups of chemical pathway(s) based on predefined parameters.

In another embodiment of the method of compactification of chemical pathway(s), the method further includes removing cycles of at least one of intermediates and transformations from the acquired plurality of chemical pathways based on the identification of at least one of the plurality of the reactions acting on same reactant and product pair, the one or more of chemical pathways formed by same set of the transformation rules, and the one or more of chemical pathways having similar intermediates.
In yet another embodiment of the present invention, a device for compactification of chemical pathway(s) for transforming a given start compound(s) to target compound(s) is disclosed. The device includes a memory, and processor(s) operatively coupled to the memory. The processors are configured to perform the steps including: (a) acquiring a plurality of chemical pathways for transforming the given start compound(s) to target compound(s), wherein the chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules based on a knowledgebase; (b) identifying at least one of a plurality of the reactions acting on same reactant and product pair where plurality of the chemical reactions being predicted using different transformation rules, and a plurality of chemical pathways, formed by same set of the transformation rules, predicted for the given one or more start compounds to one or more target compounds, and a plurality of chemical pathways, having similar intermediates, predicted for the given one or more start compounds to one or more target compounds; (c) grouping together at least one of the identified plurality of the reactions acting on the same reactant and product pair, and at least one of the one or more chemical pathways formed by the same set of transformation rules and the one or more chemical pathways having similar intermediates; and (d) compactifying the one or more groups of the one or more chemical pathways for transforming the given one or more start compounds to one or more target compounds.
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS
The aforementioned aspects and other features of the present invention will be explained in the following description, taken in conjunction with the accompanying drawings, wherein:
Figure 1 is a schematic flow representation of a method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s), according to an embodiment of the present invention.
Figure 2 is a schematic diagram for prediction of reaction steps and output molecule(s) for transforming a start compound to a target compound and also for transforming the target compound to the start compound, according an embodiment of the present invention.
Figure 3 is a schematic diagram for prediction of reaction steps and output molecule(s) for transforming multiple start compounds to a target compound and for transforming the target compound to the multiple start compounds, according an embodiment of the present invention.
Figure 4 illustrates an example of the multidirectional prediction of reaction steps, according an embodiment of the present invention.
Figure 5 is a block level diagram of a device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s), according an embodiment of the present invention.
Figure 6 is a schematic flow representation of a method for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s), according to another embodiment of the present invention.
Figure 7 depicts a system of simultaneous linear equations representing change in functional groups while moving from a start to target compound in a transformation matrix, according to the another embodiment of the present invention.
Figure 8 depicts an exemplary embodiment for solution of linear equations, according to the another embodiment of the present invention.
Figure 9 is a block level diagram of a device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s), according the another embodiment of the present invention.
Figure 10 is a schematic flow representation of a method of compactification of chemical pathway(s) for transforming a given start compound(s) to target compound(s), according to yet another embodiment of the present invention.
Figure 11a is a schematic diagram for grouping at chemical pathway level, according to yet another embodiment of the present invention.
Figure 11b depicts an example of the grouping at chemical pathway level, according to yet another embodiment of the present invention.
Figure 11c depicts an example of the compactification of groups of the chemical pathways, according to yet another embodiment of the present invention.
Figure 12 is a block level diagram of a device for compactification of chemical pathway(s) for transforming a given start compound(s) to target compound(s), according to the yet another embodiment of the present invention.
The drawings described herein are for illustration purposes only and are not intended to limit the scope of the present disclosure in any way.
DETAILED DESCRIPTION OF THE INVENTION
The embodiments of the present invention will now be described in detail with reference to the accompanying drawings. However, the present invention is not limited to the embodiments. The present invention can be modified in various forms. Thus, the embodiments of the present invention are only provided to explain more clearly the present invention to the ordinarily skilled in the art of the present invention. In the accompanying drawings, like reference numerals are used to indicate like components.
The specification may refer to “an”, “one” or “some” embodiment(s) in several locations. This does not necessarily imply that each such reference is to the same embodiment(s), or that the feature only applies to a single embodiment. Single features of different embodiments may also be combined to provide other embodiments.
As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless expressly stated otherwise. It will be further understood that the terms “includes”, “comprises”, “including” and/or “comprising” when used in this specification, specify the presence of stated features, integers, steps, operations, elements and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. As used herein, the term “and/or” includes any and all combinations and arrangements of one or more of the associated listed items.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The present invention provides methods for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s).
In an embodiment of the present invention, a method of in silico prediction of chemical pathway(s) for transforming the start compound(s) to target compound(s) by multi-directionally predicting output molecule(s), through reaction prediction steps, produced by each of given input(s) based on a knowledgebase is disclosed.
The flow diagram as given in Figure 1 provides the detailed steps of the present method, according to the embodiment.
At step 102, the process starts with performing reaction prediction steps for multi-directionally predicting output molecule(s)/intermediate(s) for a given set of start and target compound(s). The reaction prediction step(s) leads to the transformation of the start compound(s) to target compound(s) and target compound(s) to start compound(s). The output molecule(s)/intermediate(s) are predicted at each of the reaction prediction step while using a knowledgebase.
For example, as depicted in Figure 2, the input of one start compound and one target compound is received for processing. The two triangles represent the individual prediction space from the start (green triangle) and target (blue triangle) compounds. The overlapping region represents pathways that lead from start to target compound. In this scenario, the chemical pathways are predicted based on the reaction prediction steps and predicted output molecule(s) involved in transformation of the start compound to the target compound and also for transformation of the target compound to the start compound.
In another example, as depicted in Figure 3, the input of 3 start compounds and one target compound is received for processing. The triangles represent the individual prediction space from the start (green triangles) and target (blue triangle) compounds. In this scenario, the chemical pathways are predicted based on the reaction prediction steps and predicted output molecule(s) involved in transformation of each of the start compound to the target compound and also for transformation of target compound to each of the start compounds.
The knowledgebase includes list of reactions, chemical moiety(ies) present in each of the listed reaction, change taking place in the chemical moiety(ies) present in the listed reactions during the occurrence of reaction, set of transformation rule(s) governing each of the enlisted reaction, and the set of transformation rule(s) represented by a unique list of transformation(s).
The output molecule(s)/intermediate(s) are predicted at each of the reaction prediction step by applying a set of transformation rule(s), included in the knowledgebase, on the inputs. Thereafter, only those transformation rules and corresponding reaction prediction step(s) is/are recorded which bring(s) transformation in the given input. Further, the output molecule(s) produced at a previous reaction prediction step become input for next reaction prediction step.
It is to be understood that the input(s) varies as per sequence of the reaction prediction steps. For example, at initial reaction prediction step the start compound(s) and/or target compound(s) are taken as input for producing the output molecule(s). In the next reaction prediction step the output molecule(s) so produced are referred to as intermediate(s) is/are taken as input and so on i.e. recursively processed, to predict the chemical pathway(s) (Figure 4). At the final reaction prediction step, the intermediate(s) so produced is/are considered as final output(s) which could be the initial input of start compound(s) or target compound(s).
Further, number of the reaction prediction steps for the start compound(s) and the target compound(s) are preset, wherein the preset number of reaction prediction steps for the start compound(s) may vary or could be equal to the preset number of reaction prediction steps for the target compound(s). This decides the total number of reaction prediction steps to be performed for the initial input (i.e. that start compound(s) and target compound(s)).
All the predicted output molecule(s) are collected into a set of intermediate molecule(s) at step 104. This step is performed after every reaction prediction step.
Similarity between the predicted output molecule(s) within the set of one or more intermediate molecules is computed at step 106. The computation is performed after each of the reaction prediction steps or after a preset number of the reaction prediction steps. Based on the computed score of the predicted output molecule(s) the group(s) of similar output molecule(s) are created. Further to this, one output molecule is selected as representative member for each of the group(s) of the similar output molecule(s).
There are several known methods of computing the similarity score for compounds (output molecule(s)/intermediate(s) in the present case). The methods uses features such as, but not limited to, chemical fingerprints and sub-structure match. Similarity quantification between two compounds are performed by computing similarity metric such as, but not limited to Tanimoto coefficient and Jaccard score (Willett, 2013; Cereto-Massagué et al., 2015).
Figure 4 depicts an example of the multidirectional prediction of reaction steps, where one of the intermediate produced at the second reaction prediction step for transforming the start compound to the target compound matches with the intermediate produced in the first step reaction prediction step for transforming the target molecule to the start compound. The same can be applied to a scenario where n number of start compounds and n number of target compounds are received as input.
Each of the group(s) of similar molecules within the set of intermediate molecule(s) is/are replaced with the single representative member at step 108, thereby using the representative member of each of the one or more groups of similar molecules as input(s) for next reaction prediction step.
Therefore two important steps have been undertaken to reduce redundancy in pathway prediction computation and computed pathway data: First - only the similar multi-directionally predicted output molecule(s) and remaining non-similar multi-directionally predicted output molecule(s) are selected to be processed through steps 102 and 104, and Second – in case several similar multi-directionally predicting output molecules are selected then only the representative member for each of the group(s) of the similar output molecule(s) are selected to be processed through steps 102 and 104.
At step 110 it is analyzed that whether total number of reaction prediction step(s) performed on the representative member(s) within the set of intermediate molecules equals to the preset number of the reaction prediction steps from the initial input start compound(s)/target compound(s). The next step would depend on the output of the analysis. If the output of analysis presents that the total number of reaction prediction step(s) performed is equal to the preset number of the reaction prediction steps from the initial input start compound(s)/target compound(s), then process of performing steps 102 to 108 in succession is stopped and step 114 commences. In another scenario, if output of analysis presents that the total number of reaction prediction step(s) performed is not equal to the preset number of the reaction prediction steps from the initial input start compound(s)/target compound(s), then step 112 commences and process of performing steps 102 to 108 in succession is repeated again, wherein the input for the step 102 would be the representative member(s) of the groups.
At the step 112, sequential reaction prediction step(s) for the set of intermediate compounds from the initial input start/target compound(s) is recorded and the representative member(s) are selected as the input for next round of reaction prediction step(s).
The start compound(s), the target compound(s), the multi-directionally predicted one or more output molecules, and sequence of the one or more reaction prediction steps are connected together to predict the one or more chemical pathways at step 114.
The predicted set of chemical pathways include sequential arrangement of a plurality of chemical reactions (reaction prediction steps) governed by one or more transformation rules.
In the Figure 4 the similarity between the predicted output molecules within the set of intermediate molecules is computed after preset number of reaction prediction steps. The preset number is 2 steps for start to target compound, while present number is 1 step for target to start compound. During this process a similar molecule ‘g’ was found and thereafter based on that only the start compound ‘S’, intermediate molecules ‘b’ and ‘g’, and target compound ‘T’ along with the corresponding reaction prediction steps and transformation rules governing those reaction prediction steps are connected to predict a chemical pathway (i.e. S ?b?g?T) effecting the transformation of start compound to target compound. Therefore, reduces redundancy in pathway prediction computation and computed pathway data.
It is to be appreciated that as the number of predictions grows exponentially with the length of simulation, the present method significantly reduces the predicted data. For brevity, a four step one-way prediction can be transformed into a two-step bi-direction prediction as the present method. For example, a typical simulation with a rule library of ~100 rules gives 25 predictions on average for each input molecule. For simulation of length n, there will be 25¬n predictions.
Reactions predicted ~ 25n, where n is the length of simulation
Reactions predicted : (a) for one-direction simulation of length 4: 390625 reactions (prior art); (b) for bi-directional simulation (length 2 + 2): 1250 reactions. It is evident that manifold reduction in the number of reactions can be achieved by the present method.
In a scenario depicted in Figure 2, assuming that each intermediate molecule predicted in start compound to target compound simulation is also predicted in bottom-up prediction, the simulation will result in about 625 pathways. This is less than 1% of total pathways predicted in a one-directional simulation. In practice, overlap (similarity) between the intermediates will be much lesser, resulting in much smaller pathways to be assessed.
In a scenario depicted in Figure 3, the simulation proceeds from both directions as described earlier, beginning from the multiple start compounds and the target compound. The predicted reactions are then linked up to form pathways. The amount of data produced depends upon the number of start (or target) compounds used. Reactions predicted - bi-directional simulation (2 x 100 start compounds + 2): 63125 reactions. These still results in ~84% reduction in data produced and predicted pathways to be assessed.
In a further embodiment, after predicting a set of chemical pathways for the transformation of start compound(s) to the target compound(s) as discussed for step 114, the predicted chemical pathways are sorted based on reaction feasibility and/or kinetics and/or abundance of intermediates formed in the chemical pathway.
In an alternative embodiment, one or more predicted chemical pathways are selected from the sorted chemical pathways based on reaction feasibility and/or kinetics and/or abundance of intermediates formed in the chemical pathway. This gives the user most relevant and feasible chemical pathway out of the several predicted ones.
Figure 5 is a block level diagram of a device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) in accordance with an embodiment of the present invention. The device is configured to predict chemical pathway(s) for a given input.
The device 500 includes processor(s) 506, and memory 502 coupled to the processor(s) 506.
The processor(s) 506, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor, a microcontroller, a complex instruction set computing microprocessor, a reduced instruction set computing microprocessor, a very long instruction word microprocessor, an explicitly parallel instruction computing microprocessor, a digital signal processor, or any other type of processing circuit, or a combination thereof.
The memory 502 includes a plurality of modules stored in the form of executable program which instructs the processor 506 to perform the method steps illustrated in Figure 1. The memory 502 has following modules: multi-directional output prediction module 508, output collection module 510, similarity computing module 512, group replacing module 514, reaction step counting module 516, reaction step recording and input selection module 518, and chemical pathways predicting module 520.
Computer memory elements may include any suitable memory device(s) for storing data and executable program, such as read only memory, random access memory, erasable programmable read only memory, electrically erasable programmable read only memory, hard drive, removable media drive for handling memory cards and the like. Embodiments of the present subject matter may be implemented in conjunction with program modules, including functions, procedures, data structures, and application programs, for performing tasks, or defining abstract data types or low-level hardware contexts. Executable program stored on any of the above-mentioned storage media may be executable by the processor(s) 506.
The multi-directional output prediction module 508 instructs the processor(s) 506 to perform the step 102 (Figure 1).
The output collection module 510 instructs the processor(s) 506 to perform the step 104 (Figure 1).
Similarity computing module 512 instructs the processor(s) 506 to perform the step 106 (Figure 1).
Group replacing module 514 instructs the processor(s) 506 to perform the step 108 (Figure 1).
Reaction step counting module 516 instructs the processor(s) 506 to perform the step 110 (Figure 1).
Reaction step recording and input selection module 518 instructs the processor(s) 506 to perform the step 112 (Figure 1).
Chemical pathways predicting module 520 instructs the processor(s) 506 to perform the step 114 (Figure 1).
In another embodiment of the present invention, a method of in silico prediction of chemical pathway(s) for transforming the start compound(s) to target compound(s) by performing a directed search for identifying a sequence of the transformations taking place in the start compound(s) and/or intermediate(s) on application of provided transformation rule(s).
The present embodiment exploits the sequential conversion in chemical moiety(ies) of functional group(s) of the start compound(s) and/or intermediate(s) during application of the transformation rule(s). During the process, the sequence of the reaction step(s) associated with the transformation rule(s) causing the transformation in the functional group(s) of the start compound(s) and/or intermediate(s) is recorded and analyzed with help of a rule transformation matrix.
The chemical moiety is a part of a molecule that may include either whole functional group(s) or parts of functional group(s) as substructures. For example, an ester (RCOOR') has an ester functional group (COOR) and is composed of an alkoxy moiety (-OR') and an acyl moiety (RCO-). As per the definition there could be situations where moieties may contain functional groups which may contain moieties. The functional groups are specific groups of atoms or bonds within molecules that are responsible for the characteristic chemical reactions of those molecules.
The molecular transformation comprises of chemical moiety residing on a molecule and at least one of chemical bond change(s), bond rearrangement(s) and chemical state change(s) it undergoes in a reaction process.
The flow diagram as depicted in Figure 6 provides detailed steps of the present method, according to the embodiment.
The chemical moiety(ies) present in the start compound(s) and the target compound(s) are identified at step 602. The process of identification of chemical moiety(ies) is performed by method(s) known in the art. The identification is performed with reference to the knowledgebase for a pair of the one start compound and the one target compound. There could be more than one pair of start compound and the target compound, if initial input includes a plurality of start or target compounds. In such cases, one pair comprising of one start compound and one target compound, is processed at a given time for identification of the chemical moiety(ies).

For example, the input received includes ethanol as start compound and ethanoic acid (acetic acid) and ethanoyl CoA (acetyl CoA) as two target compounds. Two pairs are formed: (a) ethanol and acetic acid; and (b) ethanol and acetyl CoA. The first pair of ethanol and acetic acid is processed to identify the presence of CH group, hydroxyl (OH) and carboxylate (C(=O)OH) chemical moieties. The second pair of ethanol and acetyl CoA is processed to identify the presence of CH group, hydroxyl (OH) and CoA chemical moieties.

The identified chemical moieties pertaining to a functional group from each pair of the start and target compounds are extracted at step 604. There could be at least one functional group in one of the given start compound, intermediate molecule(s) and target compound. Taking forward the aforementioned example of the pair of ethanol and acetic acid, the following chemical moieties pertaining to functional groups are extracted, CH group, hydroxyl group (-OH) and acetic acid (C(=O)OH).
A functional chemical moiety vector representing functional groups for the start and target compound(s) received as input is constructed based on the knowledgebase at step 506. For example, a functional chemical moiety vector constructed for the extracted chemical moieties [CH group, hydroxyl group (-OH) and acetic acid (C(=O)OH)], pertaining to functional groups, is represented as following:
Functional chemical moieties Start Compound(Ethanol) Target Compound(Acetic Acid)
C-H 5 3
C-OH 1 0
C(=O)OH 0 1

Difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector the input start compound is computed at step 608.
Chemical Moieties Target Compound (Ethanol) Start Compound (Ethanol) Difference of functional moieties
C-H 3 5 -1
C-OH 1 - 1 = 0
C(=O)OH 1 0 1

Transformation vector is computed based on the computed difference between the functional chemical moiety vector of the input target compound and the functional chemical moiety vector of input compound and a rule transformation matrix at step 610. The rule transformation matrix comprises a plurality of columns, where each column represents one transformation rule, and a plurality of rows, where each row represents one functional group of the identified chemical moiety(ies) from the input. Therefore, the number of columns in the matrix is directly proportional to the number of transformation rules present in the knowledgebase or number of transformation rules selected to be used for this purpose.
Intermediate(s) is/are produced after application of a set of transformation rules (taken from the knowledgebase and enlisted in the rule transformation matrix) on the initial input of the start compound. The matrix is populated based on the effect of transformation rules on the input (explained in later part) and relevant transformation rule(s) and association reaction step is identified. This forms the first step in prediction of reaction steps. Further, the intermediate(s) produced at a previous chemical reaction prediction step become input for the next chemical reaction prediction step. The process is repeated recursively till it yields the target compound(s).
Once the difference between the functional chemical moiety vectors of the input is known a rule transformation matrix, T, is constructed wherein rows correspond to the identified functional groups and columns to transformation rules (Figure 7). The rule transformation matrix is populated based on predefined rules: (a) negative entry in the matrix denotes that the transformation rule(s) acted on the identified chemical moiety leading to either deletion or modification of the identified chemical moiety, (b) positive entry in the matrix denotes that the transformation rule(s) acted lead to the formation of a new chemical moiety, and (c) zero entry in the matrix denotes no effect brought upon the identified chemical moiety by application of the transformation rules.
The difference in functional groups between the start and target molecules is identified and recorded as vector D. To identify a set of coefficients corresponding to transformation rule applications the following formulae is used:
T·C = D
where C is the coefficient vector
This can be solved using known methods for solution of simultaneous linear equations. In one embodiment, graph traversal method is used to identify rule paths that will be needed to convert between the functional group compositions of start molecule into that of target molecule. An example of the embodiment of the ‘directed search’ method based on the knowledgebase comprising of 4 transformation rules while having functional groups ‘–CH’ and C(=O)SCoA in the start and target compound, respectively is depicted in Figure 8. The transformation rules and modification in chemical moiety on the input caused are following:
R1: C-H ? C-OH
R2: C-OH ? C=O
R3: C=O ? C(=O)OH
R4: C=O ? C(=O)SCoA
For converting CC to CC(=O)O, the method predicts use of Rules R1, R2 and R3. Therefore, the present invention helps in identification of potential start compound(s) and also the transformation rule(s) required causing the transformation. It results in significant reduction in search space exploration.
Sequence of transformations from the one or more start compounds and/or the one or more intermediates to the one or more target compounds based on the computed transformation vector is identified at step 612. Taking forward the example illustrated in Figure 8, the sequence of transformation and relevant transformation rules predicted are:
C-H C-OH C=O C(=O)OH
The chemical pathway(s) is predicted by constructing chemical pathway(s) based on the sequence of the transformations identified while applying corresponding transformation rule(s) unto the start compound(s) and intermediate(s) to predict the target compound(s) at step 614.
As discussed earlier, the chemical pathway includes sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules.
The chemical pathways so predicted could be further refined in terms of reducing the data to be assessed for finding out most efficient chemical pathway for bringing out transformation or any other user requirement. Refinement would enable user(s) to select the most appropriate pathway out of all the predicted or received chemical pathways for transforming given input(s) in a more efficient manner.
Figure 9 is a block level diagram of a device for in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) in accordance with another embodiment of the present invention. The device is configured to predict chemical pathway(s) for a given input.
The device 900 includes processor(s) 906, and memory 902 coupled to the processor(s) 906.
The processor(s) 906, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor, a microcontroller, a complex instruction set computing microprocessor, a reduced instruction set computing microprocessor, a very long instruction word microprocessor, an explicitly parallel instruction computing microprocessor, a digital signal processor, or any other type of processing circuit, or a combination thereof.
The memory 902 includes a plurality of modules stored in the form of executable program which instructs the processor 906 to perform the method steps illustrated in Figure 6. The memory 902 has following modules: chemical moiety(ies) identification module 908, chemical moiety(ies) extraction module 910, functional chemical moiety vector construction module 912, computation module 914, transformation sequence identification module 916, and chemical pathways predicting module 918.
Computer memory elements may include any suitable memory device(s) for storing data and executable program, such as read only memory, random access memory, erasable programmable read only memory, electrically erasable programmable read only memory, hard drive, removable media drive for handling memory cards and the like. Embodiments of the present subject matter may be implemented in conjunction with program modules, including functions, procedures, data structures, and application programs, for performing tasks, or defining abstract data types or low-level hardware contexts. Executable program stored on any of the above-mentioned storage media may be executable by the processor(s) 906.
The chemical moiety(ies) identification module 908 instructs the processor(s) 906 to perform the step 602 (Figure 6).
The chemical moiety(ies) extraction module 910 instructs the processor(s) 906 to perform the step 604 (Figure 6).
The functional chemical moiety vector construction module 912 instructs the processor(s) 906 to perform the step 606 (Figure 6).
The computation module 914 instructs the processor(s) 906 to perform the steps 608 and 610 (Figure 6).
The transformation sequence identification module 916 instructs the processor(s) 906 to perform the steps 612 (Figure 6).
The chemical pathways predicting module 918 instructs the processor(s) 906 to perform the steps 614 (Figure 6).
In yet another embodiment of the present invention, a method of compactification of chemical pathway(s) predicted or acquired for transforming given start compound(s) to target compound(s) is provided. The present embodiment achieves the compactification by performing grouping at level of the chemical reactions and the chemical pathways, thereby reducing redundancy in the total amount of reactions to be assessed. It is to be noted that the present method of compactification can be applied in continuation with the disclosed methods of the in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) or on a set of chemical pathways provided by some other methods/outside source to refine the chemical pathways.
The flow diagram as depicted in Figure 10 provides the detailed steps of the present method of compactification, according to the embodiment.
A plurality of chemical pathways for transforming given start compound(s) to target compound(s) is acquired at step 1002. The chemical pathways comprise sequential arrangement of a plurality of chemical reactions governed by one or more transformation rules based on a knowledgebase. As mentioned earlier, the plurality of chemical pathways acquired may be the ones predicted by the methods of in silico prediction of chemical pathway(s) for transforming start compound(s) to target compound(s) [Figures 1 and 6] or from any other method or source.
The knowledgebase includes list of reactions, one or more chemical moieties present in each of the listed reaction, change taking place in the one or more chemical moieties present in of the listed reactions during the reaction, one or more transformation rules governing each of the enlisted reaction, and the set of one or more transformation rules represented by a unique list of one or more transformations.
At step 1004, the acquired chemical pathway(s) is/are analyzed to identify:
(a) the chemical reaction(s) acting on same reactant and product pair where plurality of the reactions being predicted/generated using different transformation rules present in the knowledgebase, and/or
(b) the chemical pathway(s) for the given start compound(s) to target compound(s) , where the chemical pathway(s) are formed by same set of the transformation rules, and/or
(c) the chemical pathway(s) for the given start compound(s) to target compound(s), having similar intermediates.
The chemical reactions analyzed here are the ones present in the acquired chemical pathways. The step 1002 helps in finding out different chemical reactions which after acting on same reactant yields same product. Therefore, such chemical reactions can replace each other in the chemical pathway(s) and further to that most efficient chemical reaction out of such reactions can be chosen to be used in next steps.
The step of identification is performed by using methods known in the art. Once such chemical reactions and chemical pathways are identified, it also becomes easy to remove cycle(s) and/or futile transformation(s) present in each of the chemical pathways (explained later).
The grouping is performed at levels of reaction and/or pathway at step 1006. The identified (a) chemical reactions acting on the same reactant and product pair are grouped together, and/or (b) the plurality of chemical pathways formed by the same set of transformation rules and/or the plurality of chemical pathways having similar intermediates are grouped together.
The reaction grouping deals with grouping chemical reactions that have same reactant and product pair, but predicted using different rules. Such a pair offers same path using different chemistries and potentially a different enzyme. Though the detail of exact chemistry and enzyme is needed for engineering pathway, but it may not be as important while screening them.
Further, one chemical reaction from each group is selected as representative based on physio-chemical and/or one or more statistical properties. In this way, the representative chemical reaction can also be used to replace the other chemical reactions (less efficient) of the same group in the chemical pathways. Therefore, overall it imparts desired efficiency in the chemical pathways and significantly reduces redundancy in the total amount of reactions to be assessed. Merging such grouped chemical reactions and replacing them with a single representative reaction reduces the data to be assessed by over 30% (Table 1).
For example in case of reaction level grouping after step of identification (1004) it is found that there are 5 different chemical reactions which convert A to B, 1 chemical reaction which convert reactant B to product C and 7 chemical reactions which converts C to D. Then three groups of chemical reactions are formed here where first group includes 5 chemical reactions which convert A to B, second group includes 1 chemical reaction which converts B to C, and third group includes 7 chemical reactions which converts C to D.
Grouping at chemical pathway level is schematically represented in Figure 11a. All the three pathways lead from same start to target compound, using same set of rules applied in different order. The order of application results in different intermediate molecules. Therefore, said three chemical pathways are grouped together. These pathways will be compactified into a set for pathway assessment at step 910. Therefore, the present step helps in quick assessment/ screening of chemical pathways and has potential to highlight possible alternate routes for known pathways.
Figure 11b shows an example of one such chemical pathway grouping. Pathways A, B and C have same start and end compounds and use same transformation rules. However, the pathways A and B have similar set of transformation rules than pathway C. Therefore, two groups are formed here, first group including A and B, and second group including C only.
The group(s) of the chemical pathway(s) for transforming the given start compound(s) to target compound(s) are compactified at step 908. This step brings about the pathway compactification in groups of the chemical pathways between a given pair of start and target compounds formed by same rule set. These pathways involve same chemistries/transformations happening in different order. As the order does not matter to the outcome of the pathway, they can all be assessed as a single group. Figure 11c depicts the compactification of groups of chemical pathways detailed in the Figure 11b. First group is compactified as A+B to represent one chemical pathway. Since the second group contains only one pathway, hence no further compactification is performed.
It is generally observed that the compactification of chemical pathways results in over 80% reduction in amount of data to be assessed.
Table 1. Impact of chemical pathway assessment methods on the number of predictions
Refinement Methods Reduction in number of pathways (%)
Putrescine Adipic acid
Cycle removal 12.71 12.93
Compactification 81.37 85.41
All methods combined 88.41 92.72

It is evident from the Table 1 that an overall impact of about 90% reduction in data to be assessed is observed by application of the present method.
In a further embodiment of the present invention, the compactified groups of chemical pathways formed at step 1008 are assessed based on predefined parameters. The predefined parameters for assessing the groups of chemical pathways including, but not limited to, physio-chemical and statistical properties. The assessment on preferred parameter(s) enables users to further refine the selection criteria.
In a further embodiment of the method of compactification, in order to augment the method of compactification of chemical pathway(s) - the cycle(s) of intermediates and/or transformation(s) present in each of the chemical pathways are removed before grouping at the step 1006. Once the reactions acting on same reactant and product pair or/and the plurality of chemical pathways, formed by same set of transformation rules, predicted/acquired for the given start compound(s) to target compound(s) are identified at step 1004, it becomes easy to remove the cycle(s).
The cycle(s) of intermediate(s) is/are formed in the chemical pathways when it contains same molecule/intermediate appearing two or more times. The chemical pathways having such cycles are of no use as they comprise of transformations that are redundant. The removal of cycle(s) of intermediate(s) could result in cleaning of about 5% of pathways (Table 1).
Further as discussed, chemical pathway(s) may also contain cycle(s) of transformation(s), also referred to as futile transformation pair(s), that result in no net effect on the outcome of the pathway. The transformation rule may transform a functional group and later reverse of the same rule transforms back it to its original form. Such transformation pairs are futile and pathways involving them can be discarded. The same has been schematically represented in Figure 12a, where application of forward (F.R1) and reverse (R.R1) of rule R1 has no net impact on pathway. A pathway based only on forward of rule R2 does the same net transformation.
Figure 12b shows an example of chemical pathways having futile transformations identified from simulation. To identify such pathways, all pathways between same start and end compound are analyzed. Pathways A, B and C have same start and end compounds. Each pathway in this group is represented by the rule set involved in the pathway. The pathways B and C involve application of both forward(F) and reverse(R) of rule R1 and R2, respectively. Application of these rules results in no net effect on the outcome of pathways. Removal of futile transformations results in a clean-up of over 10% (Table 1).
Figure 13 is a block level diagram of a device for compactification of chemical pathway(s) for transforming a given start compound(s) to target compound(s) in accordance with yet another embodiment of the present invention. The device is configured to predict chemical pathway(s) for a given input.
The device 1300 includes processor(s) 1306, and memory 1302 coupled to the processor(s) 1306.
The processor(s) 1306, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor, a microcontroller, a complex instruction set computing microprocessor, a reduced instruction set computing microprocessor, a very long instruction word microprocessor, an explicitly parallel instruction computing microprocessor, a digital signal processor, or any other type of processing circuit, or a combination thereof.
The memory 1302 includes a plurality of modules stored in the form of executable program which instructs the processor 1306 to perform the method steps illustrated in Figure 10. The memory 1302 has following modules: data acquisition module 1308, identification module 1310, grouping module 1312, and compactification module 1314.
Computer memory elements may include any suitable memory device(s) for storing data and executable program, such as read only memory, random access memory, erasable programmable read only memory, electrically erasable programmable read only memory, hard drive, removable media drive for handling memory cards and the like. Embodiments of the present subject matter may be implemented in conjunction with program modules, including functions, procedures, data structures, and application programs, for performing tasks, or defining abstract data types or low-level hardware contexts. Executable program stored on any of the above-mentioned storage media may be executable by the processor(s) 1306.
The data acquisition module 1308 instructs the processor(s) 1306 to perform the step 1002 (Figure 10).
The identification module 1310 instructs the processor(s) 1306 to perform the step 1004 (Figure 10).
The grouping module 1312 instructs the processor(s) 1306 to perform the step 1006 (Figure 10).
The compactification module 1314 instructs the processor(s) 1306 to perform the step 1008 (Figure 10).
The present embodiments have been described with reference to specific example embodiments; it will be evident that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the various embodiments. Furthermore, the various devices, modules, and the like described herein may be enabled and operated using hardware circuitry, for example, complementary metal oxide semiconductor based logic circuitry, firmware, software and/or any combination of hardware, firmware, and/or software embodied in a machine readable medium.