Skip to content

Files

Latest commit

366000e · Jul 18, 2023

History

History

dx_app

Cicero (St. Jude) (DNAnexus Platform App)

stjude_cicero

This is the source code for an app that runs on the DNAnexus Platform. For more information about how to run or modify it, see https://wiki.dnanexus.com/.

Cicero works in conjunction with RNA-peg to discover gene fusions.

Cicero has 5 main steps:

  1. Fusion breakpoint selection: Candidate breakpoint regions are first identified by presence of soft-clipped reads representing partial alignments to the reference genome.
  2. Local assembly: Qualified candidate regions are selected for local assembly using read pairs of soft-clips and those of one-unmapped read.
  3. Contig mapping: The resulting contig is subjected for genome-wide mapping to remove false positives caused by RNA-seq mapping artifacts in paralogous regions or common library construction artifacts. To "rescue" gene fusions that lack soft-clipped reads, which may be caused by focal deletion, "splice-junction" reads that span multiple genes are also evaluated for their potential for gene fusion. Then, genomic mapping locations of the assembled contig or the novel junction reads are used to determine the fusion partner genes
  4. Fusion annotation: Automated frame-checking is performed to evaluate whether the contig is in-frame for both partner genes.
  5. Ranking: the ranking of each candidate was determined by the transcript allelic fraction (TAF) of fusion, extend of overlap between the contig and the partner genes, consistency of transcription orientation, reading frame derived from contig with the partner genes, and whether it matches to a known fusion or ITD event.