Pharmacogenetic drug interaction management system

Abstract

A system is disclosed for personalized medical treatment based on biomarkers.

What is claimed is: 
     
         1 . A system to provide information on a drug substance for a subject, comprising:
 a network unit to receive gene or DNA sequencing data; and   a processor running:   code to match genomic biomarker(s) from gene or DNA sequencing for a population with historical information for a population on drug structure, dosage, clinical variability and risk for adverse events for the drug substance, the computer constructing side effect features for each drug, and applying a classifier to the features to predict one or more adverse drug interactions, the computer generating one or more indicia for the drug substance; and   code to apply the indicia to the subject DNA to provide personalized medicine.   
     
     
         2 . The system of  claim 1 , wherein the biomarkers comprise at least one of: germ-line or somatic gene variants, functional deficiencies, expression changes, chromosomal abnormalities, and protein biomarkers used to select patients for treatment. 
     
     
         3 . The system of  claim 1 , comprising indicia on gene, protein or chromosomal testing, genetic testing, functional protein assays, cytogenetic studies before using the drug and an indicia about changes in efficacy, dosage or toxicity due to genetic variants, an indicia on a gene or protein involved in the metabolism or pharmacodynamics of the drug, and an indicia that the gene or protein leads to different response. 
     
     
         4 . The system of  claim 1 , comprising a code communicating with: a gene-drug interactions training dataset that includes pharmaceutical and pharmacogenomics interactions for the drug; a side effect feature database for the drug from gene sensitivity associated with the drug; a gene-drug interactions classifier that predicts adverse gene-drug interactions for drug gene pairs and a subject's genetic scan; and for each of the side effects, a Fisher's exact test to determine predicted gene-drug-drug interactions. 
     
     
         5 . The system of  claim 1 , wherein the substance is a drug, comprising code for: constructing a gene-drug interactions training dataset that includes pharmaceutical, pharmacokinetic or pharmacodynamics, and pharmacogenomics drug-drug interactions for each drug; constructing side effect features for each of the plurality of drugs from side effects associated with the plurality of drugs; running a gene-drug-drug interactions classifier that predicts adverse drug-drug interactions for drug pairs and the genetic scan; and for each of the side effects, performing a Fisher's exact test to determine predicted gene-drug-drug interactions. 
     
     
         6 . The system of  claim 5 , wherein the gene-drug-drug interactions classifier uses a classifying objective function having a smoothness constraint and a fitting constraint to render the predicted adverse gene-drug interactions. 
     
     
         7 . The system of  claim 1 , comprising code for building a genetic drug-drug interactions classifier using deep learning with a neural network, comprising propagating drug-drug interactions between different ones of the plurality of drugs on a basis that if a first drug has an interaction with a second drug and the second drug is similar to a third drug, then the first drug is considered as having interaction with the third drug. 
     
     
         8 . The system of  claim 1 , wherein the indicia comprise a bar code, a near field communication (NFC) transmission, a text, or an Internet link. 
     
     
         9 . The system of  claim 1 , wherein the indicia is readable by a smart phone, comprising code to update the indicia with changes in the subject after the consumption of the substance. 
     
     
         10 . The system of  claim 1 , comprising a gene sequencer to capture genetic data from the subject, wherein the indicia is updated with changes in the subject after the consumption of the substance. 
     
     
         11 . The system of  claim 1 , comprising a module detect the evolutionary paths of escape 
     
     
         12 . The system of  claim 1 , comprising a hidden markov model (HMM) to detect evolutionary paths of escape. 
     
     
         13 . The system of  claim 1 , comprising a module to detect mutation by comparing later sequence reads by a genetic analyzer with prior sequence reads. 
     
     
         14 . The system of  claim 1 , comprising a module to detect an amount of mutation cancer polynucleotides in a sample from a subject over time by determining a frequency of the cancer polynucleotides at a plurality of time points; determining an error range for the frequency at each of the plurality of time points; determining, between an earlier and later time point, whether error ranges (1) overlap, indicating stability of frequency, (2) an increase at the later time point outside the error range, indicating increase in frequency or (3) a decrease at the later time point outside the error range, indicating decrease in frequency. 
     
     
         15 . The system of  claim 1 , comprising a model to infer tumor phylogeny from sequencing data, wherein subclones are related to each other by an evolutionary process of acquisition of mutations. 
     
     
         16 . The system of  claim 1 , comprising a deep learning machine using deep convolutionary neural networks for detecting genetic based drug-drug interaction. 
     
     
         17 . The system of  claim 1 , comprising code to:
 acquire subject genetic scans using a gene sequencer;   identify each substance to be provided to the subject;   determine substance interactions for each drug and the genetic scan; and   provide indicia associated with each substance to warn the subject or a medical provider based on the genetic scan.   
     
     
         18 . The system of  claim 1 , comprising a module for analyzing a disease state of a subject by collecting genetic profile data on a population of tumors and original tumor treatment(s); identifying one or more evolutionary paths of escape and evolved tumor treatment(s); and based on a subject profile, predicting a probability of escape along the one or more evolutionary paths. 
     
     
         19 . The system of  claim 1 , comprising a module to recommend drug repurposing using drug-effect similarities from existing therapeutic indications of known drugs. 
     
     
         20 . The system of  claim 1 , comprising a module to predict side effects before a drug enters a clinical trial.