LociAnalysis is a tool for reference-aware, but otherwise de novo analysis of long amplicon data from SMRT Sequencing. It uses BLASR to "sort" sequence data based on which references it is most similar too, and Long Amplicon Analysis to analyze the data in each bin to generate polished and phased de novo consensus sequences.
LociAnalysis can be thought of as a wrapper around Long Amplicon Analysis to allow it to handle larger or more complex samples than it was designed for, or in effect a form of "Meta Long Amplicon Analysis". Effort has been made to perserve option names and functions between Long Amplicon Analysis and LociAnalysis to facilitate moving work between the two tools.
Alleles of clinical interest from the genes in the Major Histocompatibility Complex (MHC) range from ~3kb in length for genes in Class I to over 16kb for some alleles from Class II. Even carefully designed PCR assays that target only the active sites of the target genes and break longer loci into multiple amplicons, such as the GenDx NGSgo® HLA Sequencing kit, can still contain 13 primer pairs for amplicons from 11 different HLA genes, ranging in size from 400bp to ~6kb.
Long Amplicon Analysis, on the other hand was designed to analyze multiplexed amplicon data from a single locus or from a single closely-related gene family, with amplicons and alleles within ~1kb in length of each other. In addition, the initial clustering step was not developed with samples containing more than 5 amplicons in mind.
To handle samples of this complexity and to work around these limitations in the design of LAA, LociAnalysis was concieved. BLASR is used to sort data by sequence similarity into different bins, and Long Amplicon Analysis is run indepdently on each bin, optionally with parameters tailored to the contents of that "bin". This allows LociAnalysis scale to handle both more amplicons, and amplicons more divergent sizes and prevalence rates than LAA can handle directly,
LociAnalysis requires only two inputs:
- A SMRT Analysis v3 compliant data
- A directory of reference FASTA files, one per "bin"
Sequence "bins" for data analysis are taken from the filenames themselves, such that any sequences most similar to references in "A.fasta" or "A_gen.fasta" (the preferred IMGT filename) are assigned to bin "A" for data from the HLA-A locus.
LociAnalysis was designed to mimic the output of Long Amplicon Analysis as closely as possible, only with the prefix "amplicon_analysis" replaced with "loci_analysis"
- loci_analysis.fastq
- loci_analysis_chimeras_noise.fastq
- loci_analysis_summary.csv
- loci_analysis_subreads.csv
- SMRT Analysis >= v3.1
- virtualenv
LociAnalysis requires a local installation of SMRT Analysis containing LAA v2, and a version of BLASR that is POSIX-compliant, which means SMRT Analysis v3.1 or greater.
In addition, since LociAnalysis is an experimental tool not officially supported by PacBio, Python's "virtualenv" is required so that LociAnalysis doesn't contaminate the SMRT Analysis installation.
(1) Activate the local SMRT Analysis Environment
$ module load smrtanlysis/3.1
(2) Create a local virtualenv for LociAnalysis
$ virtualenv LociAnalysisEnv
(3) Activate the new environment
$ source LociAnalysisEnv/bin/activate
(4) Install LociAnalysis from GitHub
$ pip install git+https://github.com/bnbowman/LociAnalysis
Please direct all inquiries to the following e-mail address: Brett Bowman (bbowman@pacificbiosciences.com)