Canadian Bioinformatics Workshops www bioinformatics ca Module Title

Canadian Bioinformatics Workshops www. bioinformatics. ca

Module #: Title of Module 2

Module 4 Visual Analysis of HT-seq data Marc Fiume Informatics for High Throughput Sequencing Data June 6 -7, 2013

Learning Objectives of Module • • to appreciate the different data viz tools in genomics to know when to use a particular tool to gain more experience with genome browsers to become an expert in variation inspection – single nucleotide and structural variants • to become familiar with next-gen variant analysis tools Module 4 bioinformatics. ca

Organization • Part I (9: 00 -10: 30) – genome browsers – visualizing single nucleotide and structural variants • Part II (11: 00 -12: 30) – variant search engines – finding disease-causing genetic mutations Module 4 bioinformatics. ca

Part I : browsing HT-seq data, inspecting variants Module 4 bioinformatics. ca

Why visualize our data? Module 4 bioinformatics. ca

Anscombe’s quartet • each of these datasets has the same mean and variance Module 4 bioinformatics. ca

Preattentive processing • encoded properly, outliers are easily identified Module 4 bioinformatics. ca

Preattentive processing (video) Module 4 bioinformatics. ca

Why visualize? • the human visual system is a low-cost* and highperformance – sense maker, to identify patterns – debugger, to identify issues and outliers * compared to cost of writing, debugging, and running computational scripts Module 4 bioinformatics. ca

Visualization Tools in Genomics Module 4 bioinformatics. ca

Which tool to use? • there are over 40 different genome browsers, which to use? • depends on – task at hand – kind and size of data – data privacy Module 4 bioinformatics. ca

HT-seq Genome Browsers Integrative Genome Viewer Savant Genome Browser • task at hand : visualizing HT-seq reads, especially good for inspecting previously identified variants • kind and size of data : large BAM files, stored locally or remotely • data privacy : run on the desktop, can keep all data private Module 4 bioinformatics. ca

You might also want to try UCSC Genome Browser Trackster (part of Galaxy) • New web-technologies are being applied to make HT-seq data browsing more interactive • UCSC Genome Browser has been retrofitted to display BAM files • Trackster is a genome browser that can perform visual analytics on small windows of the genome, deploy full analysis with Galaxy Module 4 bioinformatics. ca

Savant • desktop genome browser, designed for HT-seq data – emphasis on manually inspecting single nucleotide and structural variations Module 4 bioinformatics. ca

Review: structural variation detection • covered in Module 3 • two complementary approaches: – depth of coverage (DOC) – paired end mapping (PEM) Module 4 bioinformatics. ca

PEM: small insertions donor reference Module 4 bioinformatics. ca

PEM: large insertions donor reference Module 4 bioinformatics. ca

PEM: deletions donor reference Module 4 bioinformatics. ca

PEM: inversions donor reference one read inverted when mapped Module 4 bioinformatics. ca

PEM: tandem duplications donor reference order of read mappings reversed Module 4 bioinformatics. ca

Structural Variants in Savant • Savant has a visualization mode for BAM files called “Matepair (Arc)” that is specialized for identifying structural variants using the PEM methodology • it connects the locations of paired mappings by an arc – arc height represents the mapped distance – arc color represents the relative orientation of the reads (for complex rearrangements, like inverstions) Module 4 bioinformatics. ca

Savant demo Module 4 bioinformatics. ca

Lab Time Module 4 bioinformatics. ca

We are on a Coffee Break & Networking Session Module 4 bioinformatics. ca

Canadian Bioinformatics Workshops www. bioinformatics. ca

Module #: Title of Module 28

Module 4 Visual Analysis of HT-seq data Marc Fiume Informatics for High Throughput Sequencing Data June 6 -7, 2013

Quiz for Module 4 Part I Module 4 bioinformatics. ca

Question 1 which visualization mode in Savant is best for finding SNPs? why? Module 4 bioinformatics. ca

Question 2 which visualization mode in Savant is best for finding structural variations? why? Module 4 bioinformatics. ca

Question 3 what kind of event does this image depict? e. g. chr 1: 5, 195, 017 - 5, 199, 144 Module 4 bioinformatics. ca

A: INSERTION donor reference Module 4 bioinformatics. ca

Question 4 what kind of event does this image depict? chr 1: 26, 489, 321 - 26, 490, 661 Module 4 bioinformatics. ca

A: DELETION donor reference Module 4 bioinformatics. ca

Question 5 what would a heterozygous deletion look like? chr 1: 31, 574, 172 - 31, 578, 242 Module 4 bioinformatics. ca

Question 6 what kind of event does this image depict? chr 1: 81, 659, 802 - 81, 661, 916 Module 4 bioinformatics. ca

A: Inversion donor reference one read inverted when mapped Module 4 bioinformatics. ca

Question 7 what kind of event does this image depict? chr 1: 11, 050, 416 - 11, 055, 457 Module 4 bioinformatics. ca

A: Tandem Duplication donor reference order of read mappings reversed Module 4 bioinformatics. ca

Part II : visual analytics for variants this is bonus material, covered if time permits contact mfiume@cs. toronto. edu for questions Module 4 bioinformatics. ca

Genetic Variant Analysis • finding disease-causing genetic mutation is “like trying to find a needle in a haystack needlestack” • lots of variants • many distractors – many false positives • errors in sequencing • errors in variant prediction – most true positives are not causal • not related to phenotype of interest, not damaging Module 4 bioinformatics. ca

Genetic Variant Analysis • filter variants based on quality, effect, and relevance to disease variant calling annotation Modules 1 -3 Module 4 filtration visualization Module 4. 1 bioinformatics. ca

Existing Tools • command-line is powerful but not interactive • Excel / Genome Browsers are interactive but not powerful Module 4 bioinformatics. ca

chr 1 : 102, 435, 394 – 129, 485, 349 Module 4 GO bioinformatics. ca

Med. Savant, a variant search engine Module 4 bioinformatics. ca

Med. Savant • visual analytics from variant calling to disease mutation discovery variant calling annotation filtration visualization Med. Savant Module 4 bioinformatics. ca

Med. Savant demo Module 4 bioinformatics. ca

You might also want to try Var. Sifter Golden Helix SVS (commercial) • Var. Sifter works in memory, good for small projects • this space is evolving; difficult to do a comprehensive comparison • much more commercial activity compared to genome browsers Module 4 bioinformatics. ca

We are on a Coffee Break & Networking Session Module 4 bioinformatics. ca