Method and system for predicting resistance of a disease to a therapeutic agent using a neural network

A method for predicting phenotypic resistance of Human Deficiency Virus (HIV) to a therapeutic agent comprising:

• (a) providing a neural network;
• (b) training a neural network on a training data set, whereby the training data set is generated from an HIV genotype-phenotype database, wherein each member of the training data set corresponds to a genetic mutation that correlates to a phenotypic resistance of HIV, said training being performed by
• i) propagating a training data set in a feed-forward fashion,
• ii) calculating the associated error,
• iii) back propagating the error,
• iv) adjusting the weights in the neural network,
• v) minimizing the error function by repeating the steps i), ii), iii), iv),
• vi) inputting a testing data set to ensure proper training, said testing data set comprising members that correspond to at least one genetic mutation, the presence of which correlates to a phenotypic resistance of HIV to at least one therapeutic agent, which testing data set is different from the training data set:
• (c) providing a determined HIV genetic sequence from a patient by
• i) obtaining an HIV sample from the patient,
• ii) obtaining the genetic sequence from the HIV sample;
• and
• d) predicting the phenotypic resistance of HIV to the therapeutic agent by inputting the determined genetic sequence into the trained neural network which computes the predicted phenotypic resistance of HIV to a therapeutic agent, wherein the phenotypic resistance is expressed as the fold-change in the IC50 or IC90 values of one or more therapeutic agents.