FIELD OF THE INVENTION The present invention generally relates to use of animal models of neurodegenerative disorders for identification of preclinica! indices of impaired brain function and for identification of disease-modifying therapies. BACKGROUND OF THE INVENTION Alzheimer's disease (AD) is the most common form of dementia, affecting about 10% of elderly people over the age of 65 years. Small et ai, (1997) JAMA 278: 1363-1371. As human longevity continues to increase, AD presents a growing health crisis. Advances in molecular neurosdence and the identification of biomarkers for neurodegenerative disease have enabled detailed descriptions of disease pathobtology. In particular, neuroimaging biomarkers have been used to describe neurodegenerative phenotypes during preclinical and early clinical disease stages. Accumulating evidence shows that neurodegenerative disorders, including Alzheimer's disease, are characterized by an extended period of progressive loss of neuronal function prior to presentation of overt clinical symptoms. Thus, intense interest is focused on the development of effective predinical therapies that can delay or prevent dinlcal manifestations. See DeKosky et al. (2003) Science 302: 830-834; Sllverman (2004) J. Nucl, Med. 2004 45: 594-607. Preclinical intervention to delay the onset of clinical AD by 5 years Is expected to reduce prevalence of dinteal AD by 50%. Additional delay could theoretically lead to virtual elimination of the disease. Brookmeyer et ai. (1998) Am. J. Public Health 88:1337-1342. Despite enormous interest In early treatment of neurodegenerative disease, effective therapies for preclinical AD are presently ut^nown. To identify drugs that can retard or arrest deterioration of brain functions prior to clinical manifestation, and to assess the use of existing therapies during presymptomatic disease stages, the present invention provides relevant indices of neuroimaging biomarkers in AD animal models. SUMMARY OF THE INVENTION The present invention provides methods for identification of compounds for the treatment of a neurodegenerative disorder during prectinical stages. Also provided are methods for identification of disease-modifying compounds for the treatment of a neurodegenerative disorder. The methods generally include the steps of (a) administering one or more candidate compounds to a predinical animal model of a neurodegenerative disorder, (b) assessing changes in one or more disease biomarkers in the animal model relative to measures of the one or more disease biomarkers in a control animal; and (c) selecting a candidate compound that induces a change in one or more disease biomarkers toward measures of the one or more disease biomarkers in a control animal. As one example, the disclosed methods are useful for identifying disease-modifying drugs for the treatment of Alzheimer's disease prior to symptomatic development. DETAILED DESCRIPTION OF THE INVENTION The present invention provides methods that are directed to identification of disease-modifying therapies for neurodegenerative diseases. In the case of AD, disease-modifying therapies ar» agents having a therapeutic effect prior to development of amyloid plaques and cognitive deterioration. The disclosed methods represent a new approach to AD drug discovery that focuses on predinical intervention. L Biomarkers of Neurodeqenerative Disease Diagnosis and Progression As described herein, detection of biomarkers in animal models may be used to develop indices for predmicat diagnosis of neurodegenerative disorders and for identification of drugs having therapeutic efficacy during preciinftcn vSseGse stages. In specific aspects of the Invention, imaging modalities are provided for detection of impaired brain functions in AD, The term biomarker generally refers to a characteristic, trait, or feature ttiat can be objectively measured and evaluated as an Indicator of a biological process, Biomarkers may be further described as genetic, imaging, molecular/biochemical, and clinical biomarkers. See DeKosky et a), (2003) Science 302: 830-834. These designations generally refer to methods for detection, and thus a biological condition or change may be described using one or more of the above-noted categories of biomarkers. Disease biomarkers are biomarkers that show statistically significant variance (p<0.05) as compared to a non-disease or control condition, such as a difference of at least about 2-fold when compared with a control condition, or a difference of at least about 5-fold, or at least about 10-fofd, or at least about 20-fold, or at least about 50-fold, or at least about 100-fold. Control animals used to develop imaging modalities for early disease detection are animals that do not exhibit dinical or predinical disease measures. For assessment of disease biomarkers In an animal model comprising a transgene, induced mutation, or site-directed mutation, an animal of the parent line lacking the transgene or mutation constitutes an appropriate control animal. The terms predWcal and presymptomatte are used interchangeably to refer to the condition of a subject prior to diagnosis of a neurodegenerative disease according to known criteria used in the art, i.e., prior to clinical disease manifestation. Predinical patients include patients at risk for developing a neurodegenerative disease (e.g., patients carrying a genetic mutation associated with increased risk of disease), or patients presenting indices correlated with increased probability of disease development Clirtcal AD is characterized by progressive cognitive decline associated with neuronal loss, accumulation of amyloid beta (Ap, which Includes AJ340 or Aj342, or fragments thereof) in the neuropil (amyloid plaques) and in cerebral blood vessels (amyloid angiopathy), and the presence of neurofibrillary tangles (NFT). See Lee et a). (2001) Annu. Rev. Neurosci. 24: 1121-1159, Selkoe (2001) Ptiysiol. Rev. 81 (2):741-766, AD is further defined as dementia not otherwise diagnosed as multi-infarct dementia (MID), dementia with Lewy bodies (DLB), frontotemporai dementia (Including Pick's disease), Parkinson's disease, or alcohol-related dementia (Korsakoff s syndrome), Mutations in p-amyloid precursor protein (APR), presenilin 1 (PS1), presenllin 2 (PS2), and apoliproteln E (APOE) genes are linked to familial forms of AD. See Tandon et al, (2000) Com Opfri, Neurol. 13; 377-383, Non-genetic risk factors also exist For example, patients with mild cognitive impairment (MCI) show increased incidence of developing diagnosable AD. Morris et al. (2001) Arch. Neurol. 58; 397-405; Petersen (2001) Neurology 56(9): 1133-1142. Additional risk factors indude significant hippocampal atrophy and declining performance on recent memory testing. Petersen (2003) Mild Cognitive Impairment: Aging to Alzheimer's Disease. Oxford University Press, New York. Predinical AD also encompasses patients diagnosed as probable/mild AD according to National Institute of Neurological and Communicated Disorders and Stroke Criteria, which requires a demonstration of deteriorated cognition (including memory) in two or more areas of suffident magnitude to interfere with work or social function. See McKhann et al. (1984) Neuro/ogy 34:939-944. The term preclinical animal model, as used herein to describe animal models for neurodegenerative disorders, refers to an animal model of a neurodegenerative disorder at a developmental stage prior to presentation of disease symptoms. Biomarkers for precflnlcal disease stages show comparable profiles in human patients and animal models. For example, a preclinical AD animal model is characten'zed by reduced cerebral blood flow and reduced glucose utilization prior to development of amyloid plaques and/or neuroflbrillary tangles (NFT). Representative AD animal models are described herein below. To identify indices for diagnosis of neurodegeneration, neurolma^tg biomarkers are assessed in animal models of neurodegeneration and in human <* subjects at presymptomatic stages, as described in the Examples. In particular, precMnicat decreases in glucose utilization, cerebral blood flow, and changes in metabolite levels may have diagnostic value when used alone or in combination with other neyroimaging biomarkers, or in combination with one or more genetic, molecular/biochemical, or clinical biomarkers. For example, additional measures of neuronal activity, neuronal integrity, neurochemistry / metabolite levels, gliosis, amyloid deposition, the presence of neurofibriltary tangles, and/or brain volume may be used to refine measures of AD-assodated changes during predinical and dinlcal disease progression and during post-treatment recovery. To identify indices for therapeutic monitoring, neuroimaging biomarkers are assessed in animal models of neurodegeneration, including predinical models, and in patients following drug administration. Indices for therapeutic monitoring are identified as measurable changes that correlate with significant changes in disease progression, including the likelihood of developing clinical stage disease. JA AD Animal Models Any relevant model for neurodegeneration may be used In the disclosed methods, including transgenlc animals or animals bearing naturally occurring, induced, or targeted mutations. Several AO animal models are known in the art Tg2576 transgenic mice, which overexpress a mutated form of human APP695, exhibit age-dependent elevation In Ap levels as well as neuropathologicaJ, behavioral, and metabolic impairments of AD. See Hsiao et al. (1995) Neuron 15(5): 1203-1218; Hsiao et al. (1996} Science 274(5284):99-102; Hsiao (1998) Exp. Gerontol. 33(7-8):883-889; Hoi comb (1999) Behav. Genet. 29(3): 177-185; Niwa et al. (2002) Neurobiol. DIs, 9(1):61-68; U.S. Patent No. 5,877,399. Cognitive defects in Tg2576 mice are detectable by about 4 months of age and prior to development of amyloid plaques. The onset of cognitive defects correlates with accumulation of brain Ap levels, Tg2576 mice having elevated levels of Apalso show impaired production of vascular relaxing factors by cerebral endothelial celts, an inability to maintain adequate blood flow during hypotension, disruption of activity-based enhancement of cerebral blood flow, and reduction of cerebral glucose utilization (Nlwa et al. (2002) Neumbiol, DIs, 9(1):61-68; Niwa et al. (2002) Am. J. Physiol. Heart OK. Physiol. 283(1 ):H315-23; ladecola et al. (1998>}-4cte. Neumpathol. (Bort.) 98: 9-14). PSAPP mice overexpress both mutant amyloid precursor protein and mutant presenilln 1 transgenes and show rapid disease progression. Ap deposition is detectable in circulate cortex by about 10 weeks of age. By 6 months, amyloid deposition is widespread, including deposition in the hippocampus, cortex, and other brain regions. See McGowan et at. (1999) WeuraWo/. Dte. 6(4):231-244. Additional AD animal models that may be used in the disclosed methods Include an animal having 3 transgene that encodes AFP and at least one mutoSon associated with Alzheimer's disease, for example, the Swedish mutation (lysine888-methiorfne6*8 mutated to asparaglne^-ieucine898) (U.S. Patent Nos. 6,509,515 and 6,586,656); PDAPP transgenic mice, which overexpress a minigene containing human APPV717F mutation (Games at al. {1995} Nature 373: 523-527; U.S. Patent No. 6,717,031); an animal model having a transgene that encodes a 69 to 103 amino acid carboxy-terminus portion of human APP (U.S. Patent No, 6,037,521); an animal model having a transgene encoding the carboxyt-terminal 100 amino adds of human APP (U.S. Patent Nos. 5,849,999 and 5,894,078); an animal model having a transgene encoding human APP751 and APP695 (U.S. Patent No. 5,850,003); an animal model having a transgene encoding a mutant protein product of a mutated FAD presenilin-1 (PS-1) gene and human APP695 Swedish mutation (U.S. Patent No. 5,898,094); an animal model having a gene-targeted mutated FAD preseniiin-1 (PS-1) gene and a human APP695 Swedish mutation (U.S. Patent No. 6,734,336); an animal model having a gene-targeted mutated FAD presenilin-1 (PS-1) gene, a human FAD Swedish mutation, and a humanized A8 mutation (U.S. Patent Mo. 6,734,336); an animal model designated TgCRNDS having a transgene encoding a human APP696 mutation, which further Includes K670N, M671L, and V717F mutations (U.S. Patent Publication No. 0030093822); an animal model having a transgene encoding APP770 with a mutation at position 717 (U.S. Patent No. 6,300,540); an animal model having a transgene encoding tau protein (U.S. Patent Nos. 6,593,512 and 6,664,443); an animal model having a transgene encoding human receptor for advanced glycatkm endproducts (RAGE) and also encoding human APP bearing mutations linked to familial Alzheimer's disease (U.S. Patent No. 6,563,015); a transgenic animal model that overexpresses TGF-B1, optionally in combination with expression of human APP (U.S. Patent No. 6,175,057); and animals prepared based on combinations of one or more of the foregoing mutations. An animal model Is typically a rodent animal model, however, models prepared in other relevant animals (e.g., primate, cat, rabbit, guinea pig, goat, horse, cow, efc.) are also useful in the invention. I.B. Neuroimaolna Biomarkers The term neuroimaging biomarker is used herein to refer to changes in functional brain activity, or other neural changes, mat are detectable In vivo. Neurolmaglng biomarkers may be assessed !n a subject using imaging methods such as magnetic source imaging and sdntigraphic techniques. For AD patients, including patients diagnosed as probable AD and patients at risk of developing AD, relevant markers of functional brain activities include cerebral atrophy / brain volume fag., assessed by MRI), cerebral Wood flow (CBF) or cerebral Wood volume (CSV) (e.g., assessed by MRI), oxygen uptake (e.g., assessed by 15O2 SPECT}. glucose uptake (e.g., assessed by [18F]flurodeoxyglucose (FDG) PET), levels of brain metabolites, such as W-acetylaspartate and myolnositoi (e.g., assessed by MRS), microglia activation (e.g., assessed by PK11195 (1-(2-(^torophenyl>-A^ethyl-fV-(lHme^yl-prapyl)-3^lsc^uinolin carboxamide) PET), and amyloid plaque deposition (0,9., assessed by 11C-8IP (11C-Pittsburgh Compound B) PET). While amyloid plaques and neuroflbflllary tangles are indicators of clinical AD, abnormalities in other neuroimaging blomarkers may be detected during predlnlcal stages. See e.g., Schott (2002) Proc. NaO. Acad. Sd. USA. 99(7):4703-4707 (elevated atrophy rate of medial temporal lobe structures detected at early stages of AD, and extrapolation of atrophy rates suggests that pathological atrophy occurs several years prior to symptomatic disease); Rusinek et al. Radiology 229: 691-696 (longitudinal study of patients over 6 years shows that an increased rate of atrophy of medial temporal lobe predicts future cognitive decline); Bookheimer et al. (2000) N. Engt. J. Med. 343(7):450-456 and Smith et al. (1999) Neurology 53: 1391-1396 (cerebrovascutar changes are also observed in patients with AD-linked genetic mutations prior to development of symptomatic disease); Reiman et al. (2001) Proc. Natl. Acad. Sd. U.S.A. 98(6):3334-3339; Small et al. (2000) Proc. Naff. Acad. ScL U.S.A. 97(11):6037-6042 (altered glucose utilization in AD patients at early disease stages); Cagnin et al., (2002) BUT. Neuropsychopharmacoi. 12: 581-586 (patients with minimal cognitive impairment show neuroinflammation in areas mat subsequently undergo atrcNhy prior to development of clinical dementia); Jessen et al. (2001) Neurology 57: 930-932 and Jaarsma et al. (1994) J. Neurol. Sd, 127: 230-233 (reduced levels of W-acetylaspartate in AD patients); and Parnetti et al. (1996) J. Am. Geriat. Soc. 44:133-138 (elevated levels of myo-inosttol in AD patients). I.B.1, Magnetic Source Imaging Neuroimaging of blood flow and brain volume may be performed using magnetic resonance Imaging (MRI), which creates Images based on the relative relaxation rates of water protons in unique chemical environments. As used herein, the term magnetic resonance imaging or MRI refers to magnetic source techniques aftematrvely described as one or more of conventional magnetic resonance imaging, magnetization transfer imaging (MTI), proton magnetic resonance spectroscopy {MRS}, diffusion-weighted imaging (DWI), perfusion-based imaging, and functional MR imaging (fMRi), See e.g., Rovaris et al, (2001} J. A/euro/. Scl. 186 Suppl 1:S3-9; Pomper & Port (2000) Magn. Reson. Imaging din. N. Am. 8:691-713; and references cited therein. ASL (arterial spin labeling} and CASH, (continuous arterial spin labeling) are functional magnetic resonance imaging techniques that depend on changes in inflowing blood spins that are in a different magnetic state than that of the static tissue. MR Images are sensitized by magnetically labeling blood flowing into a tissue sites of interest. This perfusion measurement is completely non-Invasive and does not require administration of contrast agents. Perfusion-wejghted images are generated by the subtraction of an Image obtained from tissue following inflowing spins from an image in which spin labeling is not performed. Perfusfon changes may be quantified by comparison to other parameters, for example, tissue T1 and the efficiency of spin labeling. CASL involves administration of a series of radiofrequency pulses, whereby blood water is repeatedly saturated. The exchange of labeled spins and brain tissue water approaches a steady state, such that the regional magnetization In the brain is directly related to cerebral blood flow. See e.g., Calamante et al. (1999} J. Cereb. Blood Flow & Metab. 19:701-735; Detre et a). (1992) Magn, Reson. Mod. 23(1): 37^15; and Floyd et al. (2003) J. Magn. Reson. Imaging 18(6): 649-655, For MRI techniques other than ASL / CASL, a contrast agent may be used to facilitate signal detection. Contrast agents for magnetic source imaging include but are not limited to paramagnetic or superparamagnetic ions, iron oxide, particles, for example monocrystaliine iron oxide nanopartide (MION) (Weissleder et al. (1992) Radiology 182(2):381-385.; Shen (1993) Magn, Reson. Mod. 29(5): 599-604) and water soluble contrast agents. Paramagnetic and superparamagnetic ions may be selected from the group of metals including iron, copper, manganese, chromium, erbium, europium, dysprosium, hoimium and gadolinium. Images derived used a magnetic source may be acquired using, for example, a superconducting quantum interference device manrwtmrteter (SQUID, available with instruction from Quantum Design of San Diego, California). See U.S. Patent No. 5,738,837. Representative methods for ASL detection of cerebral blood flow in AD animal models and in AD patients are described in Examples 3 and 7, respectively. Additional representative methods for magnetic resonance imaging of blood flow in animal models may be found in van Bruggen et at. (1998) J. Gereb. Stood Flow Metab. 18(11): 1178-1183; Mandevlle et al. {1998) Magn, Reson. Med. 39:615-624; Mueggler et al. (2001) Magn. Reson. Med. 46:292-298. Imaging of regional brain metabolism may be performed using MRS, which measures cellular activity on the basis of the levels of phospholipids, hi-energy compounds, inorganic phosphates, neurotransmitters, and amino adds. For example, energy metabolism in brain may be assessed by determining levels of adenylate and creatine phosphates, (ATP, ADP, AMP, CP), intermediates of glycolysis and the trtcarboxyiic add (TCA) cyde, TCA enzymes, oxidative phosphorylatJon, electron transfer chain complexes, and ATPases (e.g., K*-ATPase, Ca2+-ATPase). Representative methods for assessing neurochemical profiles in AD animal models and in AD patients using MRS are described in Examples 4 and 8, respectively. Additional methods for determination of metabolite levels in animal models and patients are described in Dedeoglu et al. (2004) Brain Res. 1012: 60-65 and in Sanacora et al. (2002) Am. J. Psychiatry 159:663-665. 1,6.2. Scintioraphlc imaging Scintigraphic imaging generally refers to radiolabei-based imaging and includes positron emission tomography (PET), single photon emission computed tomography (SPECT), gamma camera imaging, and rectilinear scanning. Most SPECT systems are based on the use of one or more gamma cameras that are rotated about the subject of analysis, and thus integrate radioactivity in more than one dimension. PET systems comprise an array of detectors in a ring that also detect radioactivity in multiple dimensions. A gamma camera and a rectilinear scanner each represent Instruments that detect radioactivity in a single plane. Related devices for sdntJgraphic imaging may also be used, such as a radio-imaging device that includes a plurality of sensors with animating structures having a common source focus. Scintigraphic techniques may be used for neuroimaging of AD blomarkers, including oxygen and glucose utilization, mtaugiiai acliv-uuj;;, atii: iaft-s&a** markers, such as amyloid plaques and neurofibrillary tangles (NFT). Changes in glucose utilization are detected in AD animal models and in AD patients prior to accumulation of Ap and neurofibrillary tangles. Therefore, these measures are particularly useful for developing diagnostic and therapeutic indices during predinical stages. See Niwa et al, (2002) Neurobiol. Disease 9: 61-68 and Alsop et ai. (2000) XUm, Neuroi. 47: 93-100. Other ligands may be used for scintigraphie imaging, IB,, any iigand that specifically bind to molecules involved in functional brain activity (e.g., receptors, antibodies, enzymes, and ion channels), that can be delivered to the brain in amounts sufficient for imaging, and that is rapidly cleared from normal brain tissue. Detectable labels for scintigraphic imaging include ^cobalt, ^copper, "copper, ^gaHlum 61chromium, ieeholmlum, 111indlum, 113mindium, Ia2iodine, 123lodine, 125todine, 131fod!nef 132iodine, 81mkrypton, 177lutettum, 1wmercury, 2(smercury, ^molybdenum, ^potassium, ^phosphorous, 188rhenium, 81rubidium, "rubidium, ^selenium, "selenium, ""technetium, ^thallium, mxenon, 133xenon, 169vtterbium, and ^inc. Cyclotron radioisotopes may also be used, for example, 11carbon, 13nltrogen, 15oxygen, and 18fluorine. Representative methods for neuroimaging of glucose utilization by [18F]fluordeoxyglucose (FDG) PET in AD animal models and AD patients are described in Examples 2 and 6, respectively. Neuroimaging of activated microglia may be performed as describee) by Shah et al. (1994) Nucl. Med. Biol.21: 573-581, or by Ramsay et al. (1992) lancet 339(8800): 1054-1055. PK11195 [1-<2-dilort>pherTyl)-A^mettTyl-A^(1^ethyl^ropyl)^isoo^ carboxamide), which is a Iigand for the peripheral benzodiazepine binding site abundant on phagocytic cells, Is used as a tracer motecule. Amyloid plaques may be imaged using benzothiazole amyloid-binding tracer molecules such as PIB or 3TA (2-(4* methylaminophenyl)benzothlazole). Additional tracer molecules for detection of amyloid deposits are described in Mathis et al. (2002) Bioorg. Med Chem. Left. 12: 295-298; Lee et al. (2003) Mud. Med, Biol. 30(6): 573-580; Wang et al. (2002) J. Mot. Neuroscl. 19(1-2): 11-16. AD patients show retention of the PIB amyloid tracer in areas of association cortex known to contain substantial amyloid deposits, including frontal cortex, parietal, temporal, and occipital cortex, and striatum. Klunk et at. (2004) Ann. Nowol. 55: 306-319, Neisrt^nsnfY tenglss ^ay b« vie^alizsd using a Iigand such as 1,1-dicyario-2-{6«