Analysis of transcriptomic data using similarity based modeling

Abstract

An analytic apparatus and method is provided for diagnosis, prognosis and biomarker discovery using transcriptome data such as mRNA expression levels from microarrays, proteomic data, and metabolomic data. The invention provides for model-based analysis, especially using kernel-based models, and more particularly similarity-based models. Model-derived residuals advantageously provide a unique new tool for insights into disease mechanisms. Localization of models provides for improved model efficacy. The invention is capable of extracting useful information heretofore unavailable by other methods, relating to dynamics in cellular gene regulation, regulatory networks, biological pathways and metabolism.

1 . A method for analyzing biological specimens comprising the steps of: 
 receiving multivariate observations each comprising measurements of the amounts of a plurality of constituents in biological specimens;    generating expected estimates of said amount measurements in each said observation as a function of each said observation;    generating residuals for said measurements from the difference between said estimates and said measurements for each said observation; and    testing said residuals to determine if a constituent is deviating from expected amounts in said biological specimens.    
   
   
       2 . A method according to  claim 1  wherein the step of generating expected estimates comprises employing an interpolative model to generate said estimates of an observation in response to providing the observation as input to the model.  
   
   
       3 . A method according to  claim 2  wherein said interpolative model is a kernel-based model.  
   
   
       4 . A method according to  claim 3 , wherein at least some of the biological specimens are of a first type, and said kernel-based model has been created using exemplar multivariate observations from a second type of biological specimens.  
   
   
       5 . A method according to  claim 3  wherein at least some of the biological specimens are of said second type, and said testing step comprises comparing residuals from observations from said first type of biological specimens to residuals from observations from said second type of biological specimens to determine if a constituent is deviating in said first type of biological specimens from what is expected in said second type of biological specimens.  
   
   
       6 . A method according to  claim 5  wherein said second type of biological specimens is healthy type, and said first type of biological specimens is a selected disease type.  
   
   
       7 . A method according to  claim 3  wherein said kernel-based model is a similarity-based model.  
   
   
       8 . A method according to  claim 3  wherein said kernel-based model is localized for each observation.  
   
   
       9 . A method according to  claim 3  wherein said kernel-based model uses a kernel operator that generates similarity values on an elemental basis.  
   
   
       10 . A method according to  claim 9  wherein said kernel operator generates similarity values on an elemental basis as a function of the range expected for each constituent.  
   
   
       11 . A method according to  claim 1 , further comprising the step of determining a disease diagnosis based on the results of said testing step.  
   
   
       12 . An apparatus for analyzing a biological specimen comprising: 
 a memory for storing a set of reference observations each comprised of measurements of a plurality of constituents from a biological specimen, each such observation being associated with a biological specimen, and each specimen being of a first selected class;    means for receiving input observations each comprised of measurements of the plurality of constituents from a biological specimen;    a processor for generating estimates of at least some of the measurements of the plurality of constituents in each said input observation, as a function of each input observation and as a function of said reference observations, said processor further disposed to generate residuals for at least some of those constituents for which estimates are generated, by differencing the estimates and the measurements of the input observations, said processor further disposed to test at least one of said residuals to determine if the associated constituent is deviating from expected behavior; and    means for displaying the results of the test performed by said processor.    
   
   
       13 . An apparatus according to  claim 12  wherein said processor is further disposed to aggregate residual deviations across multiple input observations corresponding to biological specimens of said first selected class; aggregate residuals deviations across multiple input observations corresponding to biological specimens of a second selected class; and compare at least one statistic of said aggregated residual deviations corresponding to at least one constituent to determine if the constituent is a candidate biomarker distinguishing said first and said second selected classes.  
   
   
       14 . An apparatus according to  claim 12  wherein said processor generates the estimates using an interpolative model based on at least some of said reference observations.  
   
   
       15 . An apparatus according to  claim 14  wherein said interpolative model is a kernel-based model.  
   
   
       16 . An apparatus according to  claim 15  wherein said kernel-based model is a similarity-based model.  
   
   
       17 . An apparatus according to  claim 16  wherein said processor is further disposed for each input observation to select which of said reference observations to base said model on by determining a subset of reference observations that are more relevant to each input observation and using the subset to generate the estimates for that input observation.  
   
   
       18 . An apparatus according to  claim 12  wherein said processor is further disposed to determine a disease diagnosis for at least one input observation based on said test.  
   
   
       19 . An apparatus according to  claim 18  wherein said processor determines a disease diagnosis for at least one input observation when a constituent exhibits a residual in that observation that has a value associated with the disease diagnosis.  
   
   
       20 . An apparatus according to  claim 18  wherein said memory stores a set of residual patterns associated with a disease diagnosis and said processor determines a disease diagnosis for at least one input observation by matching a pattern of residuals for the input observation to at least one of said set of residual patterns in said memory.