AI applications for analysis ofmulti ‘Omics’ data for identification of personalized driver pathways and Cancer therapy candidates Uğur Sezerman Acıbadem Üniversitesi
Human Genome Project Goals: ■ identify all the approximate 30,000 genes in human DNA, ■ determine the sequences of the 3 billion chemical base pairs that make up human DNA, ■ store this information in databases, ■ improve tools for data analysis, ■ transfer related technologies to the private sector, and ■ address the ethical, legal, and social issues (ELSI) that may arise from the project. Milestones: ■ 1990: Project initiated as joint effort of U.S. Department of Energy and the National Institutes of Health ■ June 2000: Completion of a working draft of the entire human genome (covers >90% of the genome to a depth of 3-4x redundant sequence) ■ February 2001: Analyses of the working draft are published ■ April 2003: HGP sequencing is completed and Project is declared finished two years ahead of schedule http://doegenomes.org http://www.sanger.ac.uk/HGP/overview.shtml U.S. Department of Energy Genome Programs, Genomics and Its Impact on Science and Society, 2003
Central Dogma of Molecular Biology
`Omics` Data
DNA Methylation Hypomethylation Hypermethylation http://www.cellscience.com/reviews7/Taylor1.jpg
5000000000000000000000000000 000
We Are Really More Bug than Man.......
GUT MICROBIOTA 10 13 - 10 14 microbes 1000- 35000 of species ( most of them are still to be identified ) Weight – 3 to 5 lbs Genome – 150 fold of our Genome Bacteroides, Prevotella , Fusobacterium , Eubacterium , Ruminococcus , Peptococc us , Peptostreptococcus , Bifidobacterium. Escherichia and Lactobacillus . Bacteroides alone constitute about 30% of all bacteria in the gut.....
Carbohydrate fermentation and absorption Digest starch, plant fiber, pectin into SCFAs (short chain fatty acids) viz. acetic acid, propionic acid, butyric acid. Digest proteins like collagen, elastin. Repression of pathogenic microbial growth Competition for nutrition, ( ruminococus and prevettella) attachment. Produce bacteriocins , Lactic acid.Also Bacillus strains produces Bacilysin which kills closteridium botullinum Metabolic function HCA (heterocyclic amines) Preventing inflammatory bowel disease SCFAs prevent IBD Preventing allergy Allergies = C. difficile and S. aureus > Bacteroides and Bifidobacteria
Low-fat, high-fiber diet Microbiota Recipient mice Increased adiposity transplant Obese twin Low-fat, high-fiber diet Lean twin Lean Low fat, High Fiber Ineffective transplant High Fat, Low Fiber Ineffective transplant Alan W Walker, Sci-Mag, Sept, 2013
Lactobacillus spp. and Bifidobacterium spp. produce GABA GABA’s natural function is to reduce the activity of the neurons to which it binds. GABA neutralizes the overexcited neurons. (anti-stress drug : Benzodiazepine)
AI/ML in Translational Medicine
AI and ML • Artificial Intelligence (AI) can be broadly defined as the science and engineering of making intelligent machines, especially intelligent computer programs • Machine Learning (ML) is an AI technique that can be used to design and train software algorithms to learn from and act https://www.fda.gov/medical-devices/software-medical- on data device-samd/artificial-intelligence-and-machine-learning- software-medical-device
ML – Major Approaches • Supervised learning – Algorithms are trained on labeled data, i.e. the desired output is known • Unsupervised learning – Algorithms are trained on unlabeled data, i.e. the desired output is unknown • Semisupervised learning, reinforcement learning, etc.
Toh TS, Dondelinger F, Wang D. Looking beyond the hype: Applied AI and machine learning in translational medicine. EBioMedicine. 2019;47:607-615.
Applications • Drug discovery – Designing chemical compounds – Drug screening • Imaging – Cell microscopy and histopathology – Radiology • Genomic medicine – Biomarker discovery Toh TS, Dondelinger F, Wang D. Looking beyond the hype: Applied AI and machine learning in translational medicine. EBioMedicine. 2019;47:607-615. – Integrating different modalities of data
Example Applications Unsupervised hierarchical clustering (part of ACME analysis) – Identified associations between BRAF mutant cell lines of the skin lineage being sensitive to the MEK inhibitör Seashore-ludlow B, Rees MG, Cheah JH, et al. Harnessing Connectivity in a Large-Scale Small-Molecule Sensitivity Dataset. Cancer Discov. 2015;5(11):1210-23. • Spectral clustering by SNF – Identification of new medulloblastoma subtypes Cavalli FMG, Remke M, Rampasek L, et al. Intertumoral Heterogeneity within Medulloblastoma Subgroups. Cancer Cell. 2017;31(6):737-754.e6. • Elastic net regression – Identification of BRAF and NRAS mutations in cell lines, were among the top predictors of drug sensitivity for a MEK Barretina J, Caponigro G, Stransky N, et al. The Cancer Cell Line inhibitor Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature. 2012;483(7391):603-7.
Zitnik M, Nguyen F, Wang B, Leskovec J, Goldenberg A, Hoffman MM. Machine Learning for Integrating Data in Biology and Medicine: Principles, Practice, and Opportunities. Inf Fusion. 2019;50:71-91.
Zitnik M, Nguyen F, Wang B, Leskovec J, Goldenberg A, Hoffman MM. Machine Learning for Integrating Data in Biology and Medicine: Principles, Practice, and Opportunities. Inf Fusion. 2019;50:71-91.
Zitnik M, Nguyen F, Wang B, Leskovec J, Goldenberg A, Hoffman MM. Machine Learning for Integrating Data in Biology and Medicine: Principles, Practice, and Opportunities. Inf Fusion. 2019;50:71-91.
Challenges Mirza B, Wang W, Wang J, Choi H, Chung NC, Ping P. Machine Learning and Integrative Analysis of Biomedical Big Data. Genes (Basel). 2019;10(2)
Challenges Mirza B, Wang W, Wang J, Choi H, Chung NC, Ping P. Machine Learning and Integrative Analysis of Biomedical Big Data. Genes (Basel). 2019;10(2)
Challenges Mirza B, Wang W, Wang J, Choi H, Chung NC, Ping P. Machine Learning and Integrative Analysis of Biomedical Big Data. Genes (Basel). 2019;10(2)
Challenges Mirza B, Wang W, Wang J, Choi H, Chung NC, Ping P. Machine Learning and Integrative Analysis of Biomedical Big Data. Genes (Basel). 2019;10(2)
Challenges Mirza B, Wang W, Wang J, Choi H, Chung NC, Ping P. Machine Learning and Integrative Analysis of Biomedical Big Data. Genes (Basel). 2019;10(2)
Our Methodology • NETWORK Based Integration of Omics Data
Our Methodology (PANOGA)
Active Subnetwork Search • Breitling et al., 2004 – mRNA expression data is used. – Significance ranks assigned to nodes. – Greedy search 𝑞 = ∏𝑗 =0 ↑𝑜 −1 ▒𝑛 − 𝑗/𝑂 − 𝑗
Partial Epilepsy Dataset # of # of # of Platform Cases Control genotyped s SNPs 3,445 6,935 528,745 SNPs Illumina, Human610- Quadv1 genotyping chips Table 5. Summary of Partial Epilepsy (PE)dataset (Kasperaviciute, et al., 2010). • 1429 patients with epilepsies of unknown cause (classified as “cryptogenic”), 919 cases with mesial temporal lobe epilepsy with hippocampal sclerosis, 241 with cortical malformations and 222 patients with various tumors, other smaller subgroups such as trauma, stroke, perinatal insults, infections, etc. • Cochran–Mantel–Haenszel test results were used as the genotypic p-values of the identified SNPs. • Using P<0.05 cutoff: • 28,450 SNPs were included.
SNP Wang CNV Rogic Previous Studies Showing Support Targeted SNPs in Epi et al. GWAS Study on et al. GWAS KEGG Term p values Genes Study OMIM on PE Epilepsy GAD Study (Aronica, et al., 2008; Okamoto, et al., 2010) Complement and coagulation cascades 2,16E-25 34 12 - Y - - - Y (Aronica, et al., 2008; Jimenez-Mateos, et al., 2008; Limviphuvadh, et al., 2010) - Cell cycle 1,03E-24 24 14 Y - - - Y Focal adhesion 7,10E-23 97 20 (Brockschmidt, et al., 2012) Y Y Y - - Y (Aronica, et al., 2008) Y ECM-receptor interaction 1,62E-22 62 14 Y - - - Y (Jimenez-Mateos, et al., 2008; Okamoto, et al., 2010) Y Jak-STAT signaling pathway 1,16E-21 24 16 Y - - - Y (Jimenez-Mateos, et al., 2008; Okamoto, et al., 2010; Zhou, et al., 2011) Y MAPK signaling pathway 2,32E-19 73 23 Y Y - Y Y (Lauren, et al., 2010) - Proteasome 1,15E-18 11 4 - - - - - (Lauren, et al., 2010) - Ribosome 1,57E-18 2 2 - - - - Y (Jimenez-Mateos, et al., 2008; Limviphuvadh, et al., 2010; Okamoto, et al., 2010; Zhou, et al., 2011) Y Calcium signaling pathway 5,73E-18 154 22 Y Y Y Y Y Regulation of actin cytoskeleton 9,23E-18 88 19 Y Y - Y - Y Adherens junction 1,01E-17 79 13 - - Y - - Y Pathways in cancer 3,94E-17 112 22 Y Y Y - - Y Gap junction 6,32E-17 147 18 (Lauren, et al., 2010) Y Y Y - - Y Apoptosis 3,72E-16 37 13 (Jimenez-Mateos, et al., 2008) Y Y - - - Y (Lauren, et al., 2010) Y Long-term depression 2,90E-15 151 15 Y Y Y Y Y (Jimenez-Mateos, et al., 2008; Limviphuvadh, et al., 2010) - Axon guidance 4,01E-15 59 12 - - - - Y Fc gamma R-mediated phagocytosis 2,22E-14 66 12 Y Y Y Y - Y Tight junction 2,82E-14 82 13 Y Y Y - - Y ErbB signaling pathway 4,04E-14 86 12 Y Y Y - - Y (Aronica, et al., 2008; Okamoto, et al., 2010) Y Wnt signaling pathway 6,28E-14 44 13 Y Y - - Y Table 6. Comparison of the top 20 SNP-targeted pathways with the pathways of the known genes, as associated to partial epilepsy.
Recommend
More recommend
