Interrogatory cell-based assays for identifying drug-induced toxicity markers

A method for identifying a modulator of drug-induced toxicity, said method comprising:

• (1) obtaining a first data set from a model for drug-induced toxicity that uses cells associated with drug-induced toxicity and represents a characteristic aspect of drug-induced toxicity, wherein the first data set represents one or more of measured genomics, lipidomics, proteomics, metabolomics, transcriptomics, and single nucleotide polymorphism (SNP) data characterizing the cells associated with drug-induced toxicity;
• (2) obtaining a second data set from the model for drug-induced toxicity, wherein the second data set represents a measured functional activity or a measured cellular response of the cells associated with drug-induced toxicity;
• (3) obtaining a third data set from comparison cells, wherein the third data set represents one or more of measured genomics, lipidomics, proteomics, metabolomics, transcriptomics, and SNP data characterizing the comparison cells;
• (4) obtaining a fourth data set from the comparison cells, wherein the fourth data set represents a measured functional activity or a measured cellular response of the comparison cells;
• (5) generating a computer-implemented first causal relationship network among the one or more of measured genomics, lipidomics, proteomics, metabolomics, transcriptomics, and SNP data and the measured functional activity or cellular response based solely on the first data set and the second data set using a programmed computing system including storage holding network model building code and a plurality of processors configured to execute the network model building code;
• wherein the generation of the first causal relationship network is not based on any known biological relationships other than the first data set and the second data set;
• and wherein generating the computer-implemented first causal relationship network comprises:
• (i) creating a list of network fragments, each network fragment including a plurality of variables connected by one or more relationships, and determining a probabilistic score associated with each network fragment based on the first data set and/or the second data set, wherein the variables correspond to the one or more of measured genomics, lipidomics, proteomics, metabolomics, transcriptomics, and SNP data and the measured functional activity or cellular response in the cells associated with drug-induced toxicity;
• (ii) creating an ensemble of trial networks, each trial network including a different subset of the list of network fragments;
• and
• (iii) globally optimizing the ensemble of trial networks by evolving the trial networks in parallel using the plurality of processors;
• (6) generating a computer-implemented second causal relationship network among the one or more of measured genomics, lipidomics, proteomics, metabolomics, transcriptomics, and SNP data and the measured functional activity or cellular response based solely on the third data set and the fourth data set using the programmed computing system, wherein the generation of the second causal relationship network is not based on any known biological relationships other than the third data set and the fourth data set;
• and
• (7) identifying, from a computer-implemented comparison of the first causal relationship network and the second causal relationship network, a causal relationship unique in drug-induced toxicity, wherein a gene, lipid, protein, metabolite, transcript, or SNP associated with the unique causal relationship is identified as a modulator of drug-induced toxicity;
• wherein the comparison of the first causal relationship and the second causal relationship includes generating a differential causal relationship network from the first causal relationship network and the second causal relationship network;
• wherein the differential causal relationship network includes one or more of:
• at least one relationship present in the first causal relationship network and absent in the second causal relationship network,
• at least one relationship present in the second causal relationship network and absent in the first causal relationship network;
• at least one relationship having a different directionality in the first causal relationship network than in the second causal relationship network;
• or
• at least one relationship having at least one significantly different parameter in the first causal relationship network than in the second causal relationship network.