Software and Datasets

Please visit our Github page for latest version of code: http://github.com/sriram-lab/

INDIGO

Drug combination design using AI/ML

1.     INDIGO (INferring Drug Interactions using chemo-Genomics and Orthology) can estimate drug interaction outcomes in pathogens such as M. tuberculosis and S. aureus, based on the degree of conservation of drug-interaction related genes in E. coli. [Download INDIGO algorithm]  (Ref: Chandrasekaran et al. Molecular Systems Biology 2016) [INDIGO Tutorial] [Download INDIGO model for Tuberculosis INDIGO-MTB]

2.     MAGENTA (Metabolism and GENomics-based Tailoring of Antibiotic regimens) is the first approach that can predict the impact of metabolic environment on the synergy of antibiotics. [Download MAGENTA software and datasets] (Ref: Cokol, Li and Chandrasekaran, Plos Computational Biology 2018) [Get latest version of MAGENTA]

3.     CARAMeL (Condition-specific Antibiotic Regimen Assessment using Mechanistic Learning) utilizes diverse datatypes and effectively searches through the large combinatorial space of both sequential and simultaneous combination therapies. [CARAMEL code] (Ref: Chung and Chandrasekaran, PNAS Nexus 2022)

4.     TACTIC (Transfer learning And Crowdsourcing to predict Therapeutic Interactions Cross-species) uses transfer learning to predict drug interactions in emerging pathogens and understudied microbes like commensals that have little or no training data. [Preprint and Code]

5.     M2D2 (Mechanistic Machine learning for Drug – Drug interactions) leverages a two-stage ML framework to predict both drug-protein and drug-drug interactions using amino acid sequences and compound structures. (Ref: Reuter et al, Biorxiv, 2024) [Preprint and Code]

6.     CALMA (Combining Antibiotics by Leveraging Metabolism-inspired Artificial neural networks) streamlines drug development by predicting both synergy and adverse toxic effects of drug combinations using neural networks. (Ref: Arora et al, NPJ Drug Discovery 2026) [CALMA Synapse page]

ASTRIX

 Systems biology models of metabolic and regulatory networks

7.     PROM (Probabilistic Regulation of Metabolism) enables the quantitative integration of regulatory and metabolic networks to build genome-scale integrated metabolic–regulatory models. [Download PROM models for E. coli and Mycobacterium tuberculosis; latest version of the PROM algorithm]   (Ref: Chandrasekaran and Price, PNAS 2010)

8.     ASTRIX (Analyzing Subsets of Transcriptional Regulators Influencing eXpression) uses gene expression data to identify regulatory interactions between transcription factors and their target genes. (Ref: Chandrasekaran et al, PNAS, 2011). [Download ASTRIX algorithm and example data]

9.     GEMINI (Gene Expression and Metabolism Integrated for Network Inference) is a network curation tool. It allows rapid assessment of regulatory interactions predicted by high-throughput approaches by integrating them with a metabolic network. [GEMINI algorithm implementation and Data]  (Ref: Chandrasekaran and Price, Plos Computational Biology 2013) [GEMINI tutorial data]

10.  CAROM (Comparative Analysis of Regulators of Metabolism) can be used for predicting the interplay between metabolism and regulation by phosphorylation and acetylation. [Source code] (Ref: Smith et al, iScience, 2022)

11.  E-GEM (EpiGenome-scale metabolic models) is the first framework for simulating the impact of cellular metabolism on a histone modification (acetylation); it accurately predicts bulk acetylation levels and impact of deacetylase inhibition based on cellular metabolic activity. (Ref: Shen et al, Genome Biology, 2019). [Download Metabolism-Epigenome model]

12.  MetOncoFit uses both genome-scale metabolic modeling and machine-learning to quantify the relative importance of various metabolic features in predicting metabolic dysregulation using gene expression, copy number variation, and patient survival data. [Supplementary website] (Ref: Oruganty et al, Cancer & Metabolism, 2020)

13.  DFA (Dynamic Flux Activity) approach infers the impact of dynamic metabolite levels on the corresponding reaction, the encompassing metabolic pathway and the entire metabolic network. DFA enabled comprehensive characterization of metabolic changes during cell-fate transitions. (Ref: Chandrasekaran et al, Cell Reports, 2017). [Download DFA source code; Tutorial data]

14.  QDS Predictor is a mechanistic metabolic model to predict impact of drugs and genetic perturbations on quiescent mammalian cells. [Github Code] (Ref: Eames & Chandrasekaran, PNAS Nexus 2024)

15.  SCOOTI uses optimization theory and machine-learning to infer cell-specific metabolic objectives and underlying metabolic trade-offs from omics data. [Github Page] (Ref: Lin et al, Cell Systems, 2025)

16.  Recon8D combines 8 different omic regulatory networks with metabolic networks to study their interplay and impact on metabolite levels. [Github Page] [Preprint] (Ref: Schildcrout et al, Biorxiv, 2025)

17.  PhosphoDDG rapidly predicts the effect of protein phosphorylation on protein stability across the entire phospho-proteome. [Github Page] (Ref: Woodard et al, Cell Reports Methods, 2025)