bamcc

Description

A set of sequence alignments (represented as a BAM/SAM/CRAM file) defines a bipartite graph in which the nodes are reference and template sequences and the edges are alignments between a template and a reference.

bamcc splits this graph into connected components and reports the mapping from reference sequences to their components.

This is useful e.g. when analyzing multireads in an RNA-Seq experiment, where connected components can be analyzed independently.

Usage

bamcc input.bam output.tsv

The output.tsv file will look like this:

seqid  seqname      component
0      FBtr0005088  0
1      FBtr0006151  1
2      FBtr0070000  2
3      FBtr0070002  0
4      FBtr0070003  2
5      FBtr0070006  3

where:

1. seqid is the 0-based number of the reference sequence. Sequence numbers are defined by the input file and are stable.
2. seqname is the reference sequence name.
3. component is the 0-based component number to which the reference has been assigned.

To extract, say, the 17th component into a separate bam file, run

samtools view -bh -o example.17.bam example.sorted.bam \
  $(awk 'BEGIN{ORS=" "} NR>1 && $3==17 {print $2}' rsem_orig.tsv)

Building

Dependencies

• A C++-14 compiler
• htslib
• Boost

Compilation

CXXFLAGS=-O2 make

This will create an executable bamcc in the current directory.

Testing

First, run make.

Then, run ./test. This will update all files test_files/example*.tsv. Failures may appear either as messages from bamcc or differences in output files reported by git diff.

Security

Do not run this program on untrusted or potentially malformed input files.