Direct classification of raw biomolecule measurement data

Abstract

Disclosed herein are systems and methods for direct classification of biological datasets. The datasets may include raw mass spectrometry data. Some aspects include training a classifier for direct classification of raw data, and some aspects include applying the classifier.

What is claimed is: 
     
         1 . A system comprising:
 a communication interface that receives data over a communication network, the data comprising datasets that separately correspond to distinct groups of biological species of one or more biological samples; and   a computer in communication with the communication interface, wherein the computer comprises one or more computer processors and computer readable medium comprising machine-executable code that, upon execution by the one or more computer processors, implements a method comprising:
 (a) combining the datasets into a multi-dimensional dataset, and 
 (b) applying a convolutional neural network to said multi-dimensional dataset to generate a label corresponding to a biological state based on said output structure, 
 wherein said convolutional neural network comprises features corresponding to m/z ratios and elution times. 
   
     
     
         2 . The system of  claim 1 , wherein said convolutional neural network generates the label based on a subset of said features. 
     
     
         3 . The system of  claim 1 , wherein said features are identified from training data sets obtained from subjects having the biological state and from subjects not having the biological state. 
     
     
         4 . The system of  claim 3 , wherein said features are identified from the training data sets by scanning the training data sets for differences in raw mass spectrometry data. 
     
     
         5 . The system of  claim 3 , wherein the subject comprises a mammal. 
     
     
         6 . The system of  claim 3 , wherein the subject comprises a human. 
     
     
         7 . A computer-implemented method of training a neural network for detection of a biological state from raw mass spectrometry data, comprising:
 collecting a first set of digital images from a database, wherein the first set of digital images comprises raw mass spectrometry data generated from subjects having a disease state;   applying one or more transformations to each digital image of the first set of digital images, including mirroring, rotating, smoothing, or contrast reduction to create a modified set of digital images;   creating a first training set comprising the collected first set of digital images, the modified set of digital images, and a second set of digital images comprising raw mass spectrometry data generated from subjects not having the disease state;   training the neural network in a first stage using the first training set;   creating a second training set for a second stage of training comprising the first training set and digital images of raw mass spectrometry data generated from subjects not having the disease state and that are incorrectly detected as generated from subjects having the disease state after the first stage of training; and   training the neural network in a second stage using the second training set.