Katana

Command-line tool to soft-clip reads from amplicon-based sequence based on specified primer locations.

The official repository is at:

https://github.com/umich-brcf-bioinf/Katana

Overview

In amplicon-based target panel sequencing, regions-of-interest are amplified by specific pairs of primers; consequently the regions-of-interest typically always start and end with these primer sequences, sequences which match the reference sequence exactly and do not reflect the actual sample sequence. In some panel designs, the amplicons may be tiled such that an amplicon of one region of interest may overlap the primer region of different amplicon. In this arrangement, the overlapping regions should enable detection of variants that fall within that primer region. However, the presence of the primer sequences will typically overwhelm the signature of true, low-frequency variants.

Katana matches each read to its corresponding primer pair based on start position of the read. Katana then soft-clips the primer region from the edge of the read sequence, rescuing the signal of true variants measured by overlapping amplicons. The output is conceptually similar to hard-clipping the primers from the original FASTQ reads based on sequence identity but with the advantage that retaining the primers during alignment improves alignment quality.

amplicon A           [ primerREGION-OF-INTERESTprimer ]
amplicon B                       [ primerREGION-OF-INTERESTprimer ]
input read1 sequence:  TGCATGAGTCTGATCTAGGTAGTTGACGTC
input read2 sequence:              ATCTAGGTAGTTGACGTCAGATAATGCAGC

output read1 sequence: tgcatgAGTCTGATCTAGGTAGTTgacgtc (clipped amplicon A primers)
output read2 sequence:             atctagGTAGTTGACGTCAGATAAtgcagc (clipped amplicon B primers)
(lowercase = soft-clipped)

Tags are added to each output read to help explain how it was modified:

• X0 : associated primer id
• X1 : original cigar string
• X2 : original reference start
• X3 : original reference_end (informational; useful for reverse reads)
• X4 : why read would be excluded (appears only if --preserve_all_alignments)

Katana assumes that:

• input bam is indexed
• primers come in sense-antisense pairs
• primer pairs are on the same chromosome
• primer chromsomes match the bam regions

• primer file is tab separated; the header line includes the following fields:

  • Customer TargetID
  • Chr
  • Sense Start
  • Antisense Start
  • Sense Sequence
  • Antisense Sequence

• primer file sense and antisense start are specified in 1-based coordinates

Quick Start

1. Install Katana (see INSTALL.rst):

$ pip install katana

2. Get the examples directory:

$ git clone https://github.com/umich-brcf-bioinf/Katana

3. Run Katana:

$ katana Katana/examples/primers.txt Katana/examples/chr10.pten.bam clipped.bam

This will read chr10.pten.bam and produce clipped.bam which contains reads adjusted to soft-clip (exclude) their respective primer regions. Unmapped reads or reads which do not match a known primer are excluded.

Katana help

$ katana --help

usage: katana primer_manifest input_bam output_bam

Match each alignment in input BAM to primer, softclipping the primer region.

positional arguments:
  primer_manifest       path to primer manifest (tab-separated text)
  input_bam             path to input BAM
  output_bam            path to output BAM


optional arguments:
  -h, --help            show this help message and exit
  -V, --version         show program's version number and exit
  --preserve_all_alignments
                        Preserve all incoming alignments (even if they are
                        unmapped, cannot be matched with primers, result in
                        invalid CIGARs, etc.)

---

Email bfx-katana@umich.edu for support and questions.

UM BRCF Bioinformatics Core