mavis.validate package

Submodules

• base module
• call module
• constants module
• evidence module
• main module

Module contents

Sub-package Documentation

The validation sub-package is responsible for pulling supporting reads from the bam file and re-calling events based on the evidence in a standard notation.

Types of Output Files

A variety of intermediate output files are given for the user. These can be used to “drill down” further into events and also for developers debugging when adding new features, etc.

expected name/suffix	file type/format	content
*.raw_evidence.bam	bam	raw evidence
*.contigs.bam	bam	aligned contigs
*.evidence.bed	bed	evidence collection window regions
*.validation-passed.bed	bed	validated event positions
*.validation-failed.tab	text/tabbed	failed events
*.validation-passed.tab	text/tabbed	validated events
*.contigs.fa	fasta	assembled contigs
*.contigs.blat_out.pslx	pslx	results from blatting contigs
*.igv.batch	IGV batch file	igv batch file

Algorithm Overview

• (For each breakpoint pair)

  • Calculate the window/region to read from the bam and collect evidence
  • Store evidence (flanking read pair, half-mapped read, spanning read, split read, compatible flanking pairs) which match the expected event type and position
  • Assemble a contig from the collected reads. see theory - assembling contigs

• Generate a fasta file containing all the contig sequences

• Align contigs to the reference genome (currently blat is used to perform this step)

• Make the final event calls

• (For each breakpoint pair)

  • call by contig
  • if fails, then call by spanning read
  • if fails, then call by split read
  • if fails, then call by mixed split read / flanking read pair
  • if fails, then call by flanking read pair. see theory - calling breakpoints by flanking evidence
  • if fails, then the event is failed

• (For each breakpoint pair)

  • determine the amount of support for the more specific call. see theory - determining flanking support

• Output new calls, evidence, contigs, etc