ighutil

Tools for working with immunoglobulin heavy chain sequences at high throughput

The project is divided into two sections: a python module for aligning sequences, and a JVM-based tool for post-processing.

Prerequisites

• samtools
• Java 7
• Python 2.7
• zlib
• A C++11-compatible compiler (tested against gcc 4.8)
• Cython (tested against v0.20)

Installation

• Run make -C clj - creates clj/bin/ighutil
• Run pip install ./python - installs vdjalign Python tool.

Alternatively, Debian packages ighutil and vdjalign can be installed from this repository:

deb     https://s3.amazonaws.com/cmccoy-debian-repo/ stable main

Packages are signed with key 3AA09EC1. Add with:

sudo apt-key adv --keyserver pgp.mit.edu --recv-keys 3AA09EC1

Then:

sudo apt-get update
sudo apt-get install -y ighutil vdjalign

Workflow

Alignment

IGH reads are aligned via the included python package, via the command ighutil align-fastq. The alignment process is simple: each sequence is aligned to all IGHV genes, keeping all alignments within a user-specified value of the maximum alignment score.

Alignment is multithreaded, and SSE3 accelerated using the excellent Smith-Waterman implementation from klib.

The remainder of the sequence is aligned against IGHJ genes; then the remainder from the best IGHJ hit against IGHD genes.

The resulting alignments are stored in BAM format.

Choosing a V/D/J assignment

Remaining post-processing uses subcommands of the JVM interface.

• ighutil reset-primary randomly assigns a primary alignment among ties for each segment, and attempts to filter out alleles not present in the data.
• ighutil calculate-match-probability calculates the expectation that each aligned base matches germline, treating all maximally scoring alignmnets as equiprobable.
• samtools view -F 260 will subset to primary alignments only.

Annotating sequences

Annotate sequences with V/D/J gene assignment, productivity status, and more using:

ighutil annotate-vdj -i <BAM_FILE> -o <CSV_FILE>