1. CSCE555 Bioinformatics Lecture 23 Integrative Genomics II Meeting: MW 4:00PM-5:15PM SWGN2A21 Instructor: Dr. Jianjun Hu Course page: http://www.scigen.org/csce555 University of South Carolina Department of Computer Science and Engineering 2008 www.cse.sc.edu. www.cse.sc.edu

2. Outline: Integrative Genomics II Integrative Microarray Analysis The Data Sources Graph mining of co-expression relationships ◦ Codense ◦ Second order relationship analysis

3. DNA RNA Protein Gene Expression Datasets: the Transcriptome Oligonucleotide Array cDNA Array Protein abundance measurement (Mass Spec) Protein interactions (yeast 2-hybrid system, protein arrays) Protein complexes (Mass Spec) Information Flow Cellular systems 1995 Bacteria ~1.6K genes 1997 Eukaryote ~6K genes 1998 Animal ~20K genes 2001 Human 30~100K genes Objective: $1000 human genome

4. Rapid accumulation of microarray data in public repositories NCBI Gene Expression Omnibus EBI Array Express 137,231 experiments 55,228 experiments The public microarray data increases by 3 folds per year

5. Multiple Microarray Technology Platforms Microarray Platforms

6. ---- Datasets Graph-based Approach for the Integrative Microarray Analysis gene experiments e g h i Annotation a b d f Functional Annotation Gene Ontology e g h i Transcriptional Annotation TF

7. Frequent Subgraph Mining Problem is hard! Problem formulation: Given n graphs, identify subgraphs which occur in at least m graphs (m  n) Our graphs are massive! (>10,000 nodes and >1 million edges) The traditional pattern growth approach (expand frequent subgraph of k edges to k+1 edges) would not work, since the time and memory requirements increase exponentially with increasing size of patterns and increasing number of networks.

8. Novel Algorithms to identify diverse frequent network patterns CoDense (Hu et al. ISMB 2005) ◦ identify frequent coherent dense subgraphs across many massive graphs Network Biclustering (Huang et al, ISMB 2007) ◦ identify frequent subgraphs across many massive graphs Network Modules (NeMo) (Yan et al. ISMB 2007) ◦ identify frequent dense vertex sets across many massive graphs

9. CODENSE: identify frequent coherent dense subgraphs across massive graphs

10. Identify frequent co-expression clusters across multiple microarray data sets c 1 c 2 … c m g 1.1.2….2 g 2.4.3….4 … c 1 c 2 … c m g 1.8.6….2 g 2.2.3….4 … c 1 c 2 … c m g 1.9.4….1 g 2.7.3….5 … c 1 c 2 … c m g 1.2.5….8 g 2.7.1….3 …...... a b c d e f g h i j k a b c d e f g h i j k a b c d e f g h i j k a b d e f g h i j k c......

11. The solution: CODENSE a novel algorithm, called CODENSE, to mine frequent coherent dense subgraphs. The target subgraphs have three characteristics: (1) All edges occur in >= k graphs (frequency) (2) All edges should exhibit correlated occurrences in the given graph set. (coherency) (3) The subgraph is dense, where density d is higher than a threshold  and d=2m/(n(n-1)) (density) m: #edges, n: #nodes

12. CODENSE: Mine coherent dense subgraph a b d e g h i c f summary graph Ĝ f a b d e g h i c G1G1 f a b c d e f g h i a b c d e f g h i a b c d e f g h i a b c d e f g h i a b c d e g h i G3G3 G2G2 G6G6 G5G5 G4G4 (1) Builds a summary graph by eliminating infrequent edges

13. (2) Identify dense subgraphs of the summary graph a b d e g h i c f summary graph Ĝ e g h i c f Sub( Ĝ) Step 2 MODES Observation: If a frequent subgraph is dense, it must be a dense subgraph in the summary graph. However, the reverse conclusion is not true. CODENSE: Mine coherent dense subgraph

14. (3) Construct the edge occurrence profiles for each dense summary subgraph e g h i c f Sub( Ĝ) Step 3 ………………… 111000e-f 011100c-i 111000c-h 111010c-f 101100c-e G6G5G4G3G2G1 E edge occurrence profiles CODENSE: Mine coherent dense subgraph

15. ………………… 111000e-f 011100c-i 111000c-h 111010c-f 111100c-e G6G5G4G3G2G1 E edge occurrence profiles Step 4 c-f c-h c-e e-h e-f f-h c-i e-i e-g g-i h-i second-order graph S g-h f-i (4) builds a second-order graph for each dense summary subgraph CODENSE: Mine coherent dense subgraph

16. c-f c-h c-e e-h e-f f-h c-i e-i e-g g-i h-i second-order graph S g-h f-i Step 4 c-f c-h c-e e-h e-f f-h e-i e-g g-i h-i Sub(S) g-h (5) Identify dense subgraphs of the second-order graph Observation: if a subgraph is coherent (its edges show high correlation in their occurrences across a graph set), then its 2nd- order graph must be dense. CODENSE: Mine coherent dense subgraph

17. (6) Identify the coherent dense subgraphs c-f c-h c-e e-h e-f f-h e-i e-g g-i h-i Sub(S) g-h Step 5 c e f h e g h i Sub(G) CODENSE: Mine coherent dense subgraph

18. CODESNE Solution The identified subgraphs by definition satisfy the three criteria: (1)All edges occur in >= k graphs (frequency) (2)All edges should exhibit correlated occurrences in the given graph set. (coherency) (3)The subgraph is dense, where density d is higher than a threshold  and d=2m/(n(n-1)) (density) m: #edges, n: #nodes

19. CODENSE: Mine coherent dense subgraph

20. CODENSE The design of CODENSE can solve the scalability issue. Instead of mining each biological network individually, CODENSE compresses the networks into two meta-graphs (the summary graph and the second-order graph) and performs clustering in these two graphs only. Thus, CODENSE can handle any large number of networks.

21. c 1 c 2 … c m g 1.1.2….2 g 2.4.3….4 … c 1 c 2 … c m g 1.8.6….2 g 2.2.3….4 … c 1 c 2 … c m g 1.9.4….1 g 2.7.3….5 … c 1 c 2 … c m g 1.2.5….8 g 2.7.1….3 … a b c d e f g h i j k a b c d e f g h i j k a b c d e f g h i j k a b d e f g h i j k c a b c d e f g h i j k a b c d e f g h i j k a b c d e f g h i j k a b d e f g h i j k c Applying CoDense to 39 yeast microarray data sets

22. Functional annotation Annotation

23. Functional Annotation (Validation) Method: leave-one-out approach - masking a known gene to be unknown, and assign its function based on the other genes in the subgraph pattern. Functional categories: 166 functional categories at GO level at least 6 Results: 448 predictions with accuracy of 50%

24. Functional Annotation (Prediction) We made functional predictions for 169 genes, covering a wide range of functional categories, e.g. amino acid biosynthesis, ATP biosynthesis, ribosome biogenesis, vitamin biosynthesis, etc. A significant number of our predictions can be supported by literature.

25. Functional annotation Annotation We made functional predictions for 779 known and 116 unknown genes by random forest classification with 71% accuracy. Variables for random forest classification: functional enrichment P-valuenetwork topology score network connectivitypattern recurrence numbers average node degreeunknown gene ratio Network size

26. Reconstruct transcriptional cascades by second-order analysis Zhou et al. Nature Biotech 2005

27. Frequently occurring tight clusters

28. Transcription Factors

29. Co-occurrence of tight clusters Coexpression network constructed with the dataset 1

30. Co-occurrence of tight clusters Coexpression network constructed with the dataset 2

31. Co-occurrence of tight clusters Coexpression network constructed with the dataset 3

32. Co-occurrence of tight clusters Coexpression network constructed with the dataset 4

33. Co-occurrence of tight clusters Coexpression network constructed with the dataset 5

34. Coexpression Networks Transcription Factors Set 1 Transcription Factor Set 2 Cooperativity

35. Relevance Networks

36. Three types of transcription cascades TF1 TF2 TF3 Type I Type II gene 2 gene 1 gene 3 gene 4 gene 5 TF1 TF2 gene 2 gene 1 gene 3 gene 4 gene 5 Type III TF1 TF2 gene 2 gene 1 gene 3 gene 4 gene 5 Transcription Regulation Protein Interaction

37. Applying to 39 yeast microarray data sets We identified 60 transcription modules. Among them, we found 34 pairs that showed high 2nd- order correlation. A significant portion (29%, p- value<10-5 by Monte Carlo simulation ) of those modules pairs are participants in transcription cascades: 2 pairs in Type I, 8 pairs in Type II, and 3 pairs in type III cascades. In fact, these transcription cascades inter-connect into a partial cellular regulatory network.

38. CDC45, RAD27, RNR1, SPT21, YPL267W YBL113C YRF1- 1 YRF1-7 SWI5 YEL077C YRF1-3 YRF1-7 ALK1 BUD4 CDC20 CDC5 CLB2 HST3 SWI5 YIL158W YJL051W YOR315W MSN4 NDD1 YBL113C YIL177C YJL225C YLR464W YML133C YRF1-1 YRF1-3 YRF1-4 YRF1-5 YRF1- 6 YRF1-7 BUD19 RPL13A RPL13B RPL16A RPL24B RPL28 RPL2B RPL33B RPL39 RPL42B RPS16B RPS18A RPS21B RPS22A RPS23B RPS4B RPS6A MET4 GCN4 MET10 MET14 MET28 SUL2 ALD5 ARG1 ARG2 ARG3 ARG4 ARG5,6 ARO1 ARO3 ARO4 ASN1 BNA1 CPA2 DED81 ECM40 HIS1 HIS4 HOM3 LEU4 LEU9 LYS20 MET22 ORT1 PCL5 TEA1 TRP2 YBR043C YDR341C YHM1 YHR162W YJL200C YJR111C Leu3 BAT1 ILV2 LEU9 SWI4 MBP1 CLB6 RNR1 SPT21 YPL267W CDC21 CDC45 CLB5 CLB6 CLN1 GIN4 IRR1 MCD1 MSH6 PDS5 RAD27 RNR1 SPO16 SPT21 SWE1 TOF1 YBR070C HGH1 LOC1 NOC3 YDR152W YHL013C YBL113C YRF1-1 YRF1-3 YRF1-7 SWI6 RCS1 GAT3 YAP5 PDR1 GAL4 RGM1 GAS1 HTA1 HTB1 Regulation of transcription modules by transcription factors, based on ChIP-chip data and supported by recurrent expression clusters Regulation between transcription factors, based on ChIP-chip data and supported by 2nd-order expression correlation Protein interactions between two transcription factors, based on experimental data and supported by 2nd-order expression correlation Two transcription modules with high 2nd-order expression correlations

39. Integrative Array Analyzer (iArray): a software package for cross- platform and cross-species microarray analysis Bioinformatics. 22(13):1665-7

40. Acknowledgement This slides are adapted from the following presentation: Xianghong Jasmine Zhou, Molecular and Computational Biology University of Southern California. Functions, Networks, and Phenotypes by Integrative Genomics Analysis IMS, NUS, Singapore, July 18, 2007