Complementing Computation with Visualization in Genomics
2 likes839 views
The document discusses the integration of computation and visualization in genomics, particularly focusing on the use of the software 'abyss-explorer' for genome assembly visualization. It highlights the processes for genome sequencing, resolving ambiguities through paired-end reads, and the effectiveness of clustering for revealing patterns in genomic data. Future work aims to enhance the tool's functionality, including search capabilities and richer analysis features.
Complementing Computation with Visualization in Genomics
- 1. British Columbia Cancer AgencyGenome Sciences CentreVancouver . British Columbia . CanadaComplementing Computation with Visualization in GenomicsMarch 11, 2010EBI Interfaces Interest ForumCydney Nielsen
- 2. Discovery pathBiological SampleGenomic DataScientific Insight
- 3. Discovery pathBiological SampleGenomic DataScientific Insight
- 4. Components of Data AnalysisAutomationAnalysisGenomic DataScientific InsightHuman Judgment
- 5. OutlineGenome Assembly VisualizationABySS-ExplorerComplement to genome browsing Using clustering and interactive data exploration
- 6. OutlineGenome Assembly VisualizationABySS-ExplorerComplement to genome browsing Using clustering and interactive data exploration
- 7. Genome Sequencingcell populationextracted DNAShotgun approachsheared DNAsequencing readsAGCGGATTGCATGACAGTGTACAGCCTGACAGAAGCGCGCTACGATCAGATCAACATGACAGTCCGAGTACATTCAGAATGGTACAGCAG
- 8. ABySS – Assembly ByShort SequencesSimpson et al. Genome Res 2009Sequencing read set (read length = 7 nt):GGACATCGGACAGACorresponding de Bruijn graph (k = 5 nt):
- 9. ABySS – Assembly ByShort SequencesSimpson et al. Genome Res 2009Sequencing read set (read length = 7 nt):GGACATCGGACAGACorresponding de Bruijn graph (k = 5 nt):ABySS merges unambiguously connected vertices to form contigs
- 10. Assembly AmbiguitiesTrue genome sequenceGGATTGAAAAAAAAAAAAAAAAGTAGCACGAATATACATAGAAAAAAAAAAAAAAAAATTACG
- 11. Assembly AmbiguitiesTrue genome sequenceGGATTGAAAAAAAAAAAAAAAAGTAGCACGAATATACATAGAAAAAAAAAAAAAAAAATTACGAssembled sequence de Bruijn graph representation
- 13. Example of existing tools: Consed
- 14. Example of existing tools: Consed
- 24. Capture sequence lengthone oscillation = 100 nt
- 25. Genome Sequencingcell populationextracted DNAread pair informationreadsheared DNAdsDNAfragment(known size)sequencing reads(typically produce millions)AGCGGATTGCATGACAGTreadGTACAGCCTGACAGAAGCGCGCTACGATCAGATCAACATGACAGTCCGAGTACATTCAGAATGGTACAGCAG
- 26. Capture read pair informationAfter building the initial single-end (SE) contigs from k-mer sequences, ABySS uses paired-end reads to resolve ambiguities.
- 27. Capture read pair informationPaired end read information is used the construct paired end (PE) contigs… 13+ 44- 46+ 4+ 79+ 70+ …blue gradient = paired end contigorange = selected single end contig
- 29. contig adjacency information
- 30. contig strand
- 31. contig length
- 34. Implemented using the Java Universal Network/Graph Framework (JUNG)
- 35. Applied the Kamada-Kawai layout algorithm (JUNG implementation)
- 36. Use ABySS files as input (version 1.1.0 and higher)
- 38. Part 1: Conclusions and Future Work Graph encoding provides a integrated display of genome assemblies and associated meta-data
- 39. This representation is particularly powerful for revealing high-level genome assembly structure, not readily viewable in any other interactive tool
- 40. Future work includes:
- 41. support for other assembly algorithm outputs
- 42. enable flexible annotation display
- 43. integrate with existing assembly editing toolsOutlineGenome Assembly VisualizationABySS-ExplorerComplement to genome browsing Using clustering and interactive data exploration
- 44. Genome Sequencingcell populationextracted DNAsheared DNAsequencing reads(typically produce millions)AGCGGATTGCATGACAGTGTACAGCCTGACAGAAGCGCGCTACGATCAGATCAACATGACAGTCCGAGTACATTCAGAATGGTACAGCAG
- 45. Genome Sequencingcell populationextracted DNAsheared DNAsequencing reads(typically produce millions)AGCGGATTGCATGACAGTGTACAGCCTGACAGAAGCGCGCTACGATCAGATCAACATGACAGTCCGAGTACATTCAGAATGGTACAGCAG
- 46. Genome Sequencingcell populationChromatin Immunoprecipitationand Sequencing (ChIP-Seq)extracted DNAselectionsheared DNAsequencing reads(typically produce millions)AGCGGATTGCATGACAGTGTACAGCCTGACAGAAGCGCGCTACGATCAGATCAAGTACAGCCTGACAGAAGCCATGACAGTCCGAGTACATTCAGAATGGTACAGCAGTTCAGAATGGTACAGCAG
- 47. Align sequences to the genomeCCGAGTACAGCCTGACAGAGCATGACAGTCCGAGTACTTGCATGACAGTCCGAGTAGCGGATTGCATGACAGTAGCGGATTGCATGACAGTAGCGGATTGCATGACAGTReference GenomeAGCGGATTGCATGACAGTCCGAGTACAGCCTGACAGARead coverageGenomic coordinate
- 48. Genome browser can reveal local patternsH3K4me3H3K36me3H3K27me3H3K9me3H3K9AcMRE
- 49. Difficult to get global overview
- 50. Focus on regions of interest1. For example, transcriptional start sites (TSS +/- 3000 nt)H3K4me3H3K9AcH3K4me1H3K36me3MeDIPMRE2. Extract data matricesNormalization for bin i, sample h:3. Cluster matrices (k-means clustering with Euclidean distance)
- 51. Focus on regions of interest1. For example, transcriptional start sites (TSS +/- 3000 nt)H3K4me3H3K9AcH3K4me1H3K36me3MeDIPMRE2. Extract data matricesNormalization for bin i, sample h:3. Cluster matrices (k-means clustering with Euclidean distance)
- 52. Focus on regions of interest1. For example, transcriptional start sites (TSS +/- 3000 nt)H3K4me3H3K9AcH3K4me1H3K36me3MeDIPMRE2. Extract data matricesNormalization for bin i, sample h:3. Cluster matrices (k-means clustering with Euclidean distance)
- 53. Enable interactive exploration4. Interactive cluster visualization (data from H1 cells)cluster size indicator (total n= 15,618)H3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAcluster (average values displayed)individual TSSH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAHOXC12 genescroll bar to explore all cluster members5. Link-out to UCSC genome browser
- 54. Enable interactive exploration4. Interactive cluster visualization (data from H1 cells)cluster size indicator (total n= 15,618)H3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAcluster (average values displayed)individual TSSH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAscroll bar to explore all cluster members
- 55. Enable interactive exploration4. Interactive cluster visualization (data from H1 cells)cluster size indicator (total n= 15,618)H3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAcluster (average values displayed)individual TSSH3K4me3H3K9AcH3K4me1H3K36me3H3K27me3H3K9me3MeDIPMREmRNAHOXC12 genescroll bar to explore all cluster members5. Link-out to UCSC genome browser
- 57. Part 2: Conclusions and Future Work Clustering reveals patterns that were not obvious using a genome browser.
- 58. Access to both global and detailed view is valuable
- 59. Future work includes:
- 60. search functionality (e.g. by region id)
- 61. integration with other clustering tools
- 62. richer analysis functionality (e.g. interactive clustering)AcknowledgementsNIH Epigenomics RoadmapABySS-ExplorerJoe Costello, UCSFPeggy Farnham, UC DavisThea Tlsty, UCSFMarco MarraMartin HirstYongjun ZhaoNina ThiessenRichard VarholShaun Jackmanİnanç BirolJason ChangLymphoma Project AnalystKaren MungallSupervisorPrimary Data GenerationSteven JonesLymphoma Genomics Team
- 65. Complementing Computation with Visualization in GenomicsMarch 11, 2010Cydney NielsenBC Cancer AgencyGenome Sciences CentreVancouver, Canada