Discovery of drug mode of action and drug repositioning from transcriptional responses Francesco Iorioa,b, Roberta Bosottic, Emanuela Scacheric, Vincenzo Belcastroa, Pratibha Mithbaokara, Rosa Ferrieroa, Loredana Murinob, Roberto Tagliaferrib, Nicola Brunetti-Pierria,d, Antonella Isacchic,1, and Diego di Bernardoa,e,1 aTeleThon Institute of Genetics and Medicine, Naples, Italy; cDepartment of Biotechnology, Nerviano Medical Sciences, Milan, Italy; eDepartment of Systems and Computer Science, “Federico II” University of Naples, Naples, Italy; dDepartment of Pediatrics, “Federico II” University of Naples, Naples, Italy; and bDepartment of Mathematics and Computer Science, University of Salerno, Salerno, Italy Presenter: Chifeng Ma Computational Genomics and Proteomics Lab Structure • Background • Method & Result • Conclusion Computational Genomics and Proteomics Lab Background Goal & Key point Drug Mode of Action New drug therapeutic effects /known Drug reposition Drug Signature Extraction Drug Distance Assessment Drug Mode of Action Construction Computational Genomics and Proteomics Lab Background Data:Connectivity Map Computational Genomics and Proteomics Lab Background cMap Data Data size: 22277*6836 Drug treated sample Gene • 1,267 compounds • several dosages • 5 cell lines: HL60, PC3, SKMEL5, and MCF7/ssMCF7 Log fold change: Log2(drug treated/normal) Computational Genomics and Proteomics Lab Method & Result Overview Computational Genomics and Proteomics Lab Method & Result Drug Signature Extraction Notation Initialization • • • • D: the set of all the possible permutations of microarray probe-set identifiers (MPI); X: a set of ranked lists of probe-set identifiers computed by sorting, in decreasing order, the genome-wide differential expression profiles obtained by treating cell lines with the same drug; δ: D2 → N: the Spearman’s Footrule distance associating to each pair of ranked lists in X, a natural number quantifying the similarity between them; B: D2 → D: the Borda Merging Function associating to each pair of ranked lists in X a new ranked list obtained by merging them with the Borda Merging Method; Computational Genomics and Proteomics Lab Method & Result Drug Signature Extraction Spearman’s Footrule Spearman’s Footrule between two samples x and y Number of genes in the sample here m=22283 The rank list place of the ith gene Computational Genomics and Proteomics Lab Method & Result Drug Signature Extraction Borda Merging Function A new ranked list of probes z is obtained by sorting them according to their values in P in increasing order Computational Genomics and Proteomics Lab Method & Result Drug Signature Extraction Prototype Ranked List Generation Once a PRL had been obtained, a signature {p,q} was extracted as the top 250 and bottom 250 as the signature. Computational Genomics and Proteomics Lab Method & Result Drug Distance Assessment Core distance algorithm: Gene Set Enrichment Analysis(GSEA) Computational Genomics and Proteomics Lab Method & Result Drug Mode of Action Construction Distance threshold Computational Genomics and Proteomics Lab Method & Result Drug Mode of Action Construction Community Identification Affinity propagation algorithm • A community is defined as a group of nodes densely interconnected with each other and with fewer connections to nodes outside the group 106 community 1309 nodes 41047 edges (856086 edges total) Computational Genomics and Proteomics Lab Method & Result Drug Mode of Action Construction Computational Genomics and Proteomics Lab Method & Result Drug Mode of Action Construction Community-Mode of Action relationship assessment • Anatomical Therapeutic Chemical (ATC) code --- 49/92 assessable communities significantly enrichment • GO enrichment analysis • MoA-Community assessment Computational Genomics and Proteomics Lab Method & Result Drug Distance Assessment Drug to Community distance Distance between Drug d and drug x Number of drugs in C which has a significant edges with drug d Computational Genomics and Proteomics Lab Method & Result Drug Net (DN) HSP90 inhibitors test • n.28 is closest, composed by the HSP90 in cMap data • n.40 n.63 Na+∕K+ATPaproteasome inhibitors n.104 NF-kB inhibitors • Computational Genomics and Proteomics Lab Method & Result Drug Net (DN) Test of cycin-dependent kinases(CDKs) inhibitors and Topoisomerase inhibitors Biology experiment was conduct to confirm that TDK inhibitors and Topo inhibitors share the universal inhibitor p21 Computational Genomics and Proteomics Lab Method & Result Drug Net (DN) • Search DN for drugs similar to 2-deoxy-Dglucose(2DOG) ---n.1---induce autophagy • Closest Drug--- Fasudil--- never been previously linked to autophagy • Biology experiment to confirm that Computational Genomics and Proteomics Lab Conclusion • Developed a general procedure to predict the molecular effects and MoA of new compounds, and to find previously unrecognized applications of wellknown drugs • Analyzed the resulting network to identify communities of drugs with similar MoA and to determine the biological pathways perturbed by these compounds. • In addition, experimentally verified a prediction • A website tool was implemented at http://mantra.tigem.it Computational Genomics and Proteomics Lab Computational Genomics and Proteomics Lab Reference • • • • • • • • • • 1. Terstappen GC, Schlupen C, Raggiaschi R, Gaviraghi G (2007) Target deconvolutionstrategies in drug discovery. Nat Rev Drug Discov 6:891–903. 2. di Bernardo D, et al. (2005) Chemogenomic profiling on a genome-wide scale using reverse-engineered gene networks. Nat Biotechnol 23:377–383. 3. Ambesi-Impiombato A, di Bernardo D (2006) Computational biology and drug discovery: From singletTarget to network drugs. Curr Bioinform 1:3–13. 4. Berger SI, Iyengar R (2009) Network analyses in systems pharmacology. Bioinformatics 25:2466–2472. 5. Hopkins AL (2008) Network pharmacology: The next paradigm in drug discovery. Nat Chem Biol 4:682–690. 6. Mani KM, et al. (2008) A systems biology approach to prediction of oncogenes and molecular perturbation targets in B-cell lymphomas. Mol Syst Biol 4:169. 7. Gardner TS, di Bernardo D, Lorenz D, Collins JJ (2003) Inferring genetic networks and identifying compound mode of action via expression profiling. Science 301:102–105. 8. Hu G, Agarwal P (2009) Human disease-drug network based on genomic expression profiles. PloS One 4(8):e6536. 9. Hughes TR, et al. (2000) Functional discovery via a compendium of expression profiles.Cell 102(1):109– 126. 10. Kohanski MA, Dwyer DJ, Wierzbowski J, Cottarel G, Collins JJ (2008) Mistranslation of membrane proteins and two-component system activation trigger antibioticmediated cell death. Cell 135(4):679–690. Computational Genomics and Proteomics Lab The End Thank you! Question? Computational Genomics and Proteomics Lab