Home

Welcome to the fusebench wiki!

What We're Thinking of Building

Features

A tool that is capable of:

• Importing results from different fusion callers tools (or can convert the results of a set of tools into a common format)
  • Initial idea: have team members write simple parsers to convert tool output formats to a common standard (most likely BEDPE)
• From the common file format, aggregating results from different tools into a consensus set
  • Initial idea: require some level of minimum overlap between features
  • Handling duplicate or 'synonym' calls might be difficult here
• Augmenting the consensus set with additional information from the RNA-Seq data set or existing tools
  • Initial idea: add gene expression data for candidate fusion partners
  • Initial idea: use Oncofuse?
• Importing information from existing databases
  • Initial idea: start from dump of CIViC, filtered for fusions
  • Initial idea: use ReCount to provide view of how common the fusion junction is in different data sets
• Review interface
  • Some kind of dashboard or web view for 'reviewers' to view evidence associated with particular fusions
• Visualization
  • TBD

Implementation

• Currently, this repository is set up as an R package. The initial idea was to implement most of the components principally in 'tidyverse'-style R, but this is not a hard requirement (it might not make sense for some parts)

Initial Data Sets

• Synthetic test set containing 9 known fusions (from FusionCatcher)
• Re-called data from publicly available data sets with known positives
• TBC - AML cell line RNA-Seq data

Component and Similar Projects

Recent Review Papers

• Comprehensive evaluation of fusion transcript detection algorithms and a meta-caller to combine top performing methods in paired-end RNA-seq data - PMID: 26582927
• Comparative assessment of methods for the fusion transcripts detection from RNA-Seq data. - PMID: 26862001

Fusion Callers

Selected tools include:

• trans-ABySS
• deFuse
• FusionCatcher
• EricScript
• pizzly
• Bellerophontes
• InFusion
• JAFFA
• FusionMap
• PRADA
• SOAPfuse
• STAR-FUSION

Visualization

• IGV
• Ribbon
• Packages like chimeraviz also have visualizations
• Packages like circlize are adaptable for this (as well as other Bioconductor packages)
• svviz - Read visualizer for structural variants

Annotation

• Computational inference:
  • OncoFuse
    • 'Oncofuse is a framework designed to estimate the oncogenic potential of de-novo discovered gene fusions. It uses several hallmark features and employs a bayesian classifier to provide the probability of a given gene fusion being a driver mutation.'
    • PMID: 23956304
• Databases:
  • CIViC
  • Database of Genomic Variants
  • Atlas of Genetics and Cytogenetics in Oncology and Haematology
  • Tumor Fusion Gene Data Portal
  • Recount
    • Note that this is actually an interface for downloading fusion junctions from many SRA studies

Aggregation

• MetaSV
  • Bioinformatics Article
  • Mainly focused on large-scale structural variants, from tools like Pindel, BreakDancer, CNVnator, BreakSeq2
• confFuse
  • GitHub link
  • Frontiers in Genetics Article
• chimeraviz
  • Bioinformatics Application Note
  • BioConductor
  • chimeraviz is able to import fusion results from 'nine different fusion-finder tools: deFuse, EricScript, InFusion, JAFFA, FusionCatcher, FusionMap, PRADA, SOAPfuse and STAR-FUSION'
  • These are imported as R S4 objects. There are then a selection of functions for visualizing and comparing events (from what I can tell the comparison is mostly done at the coordinate level)
• MAVIS - BCGSC unpublished tool along similar lines