PBHoover and CigarRoller: a method for confident haploid variant calling on Pacific Biosciences data and its application to heterogeneous population analysis

Abstract

AbstractMotivationSingle Molecule Real-Time (SMRT) sequencing has important and underutilized advantages that amplification-based platforms lack. Lack of systematic error (e.g. GC-bias), completede novoassembly (including large repetitive regions) without scaffolding, can be mentioned. SMRT sequencing, however suffers from high random error rate and low sequencing depth (older chemistries). Here, we introduce PBHoover, software that uses a heuristic calling algorithm in order to make base calls with high certainty in low coverage regions. This software is also capable of mixed population detection with high sensitivity. PBHoover’s CigarRoller attachment improves sequencing depth in low-coverage regions through CIGAR-string correction.ResultsWe tested both modules on 348M.tuberculosisclinical isolates sequenced on C1 or C2 chemistries. On average, CigarRoller improved percentage of usable read count from 68.9% to 99.98% in C1 runs and from 50% to 99% in C2 runs. Using the greater depth provided by CigarRoller, PBHoover was able to make base and variant calls 99.95% concordant with Sanger calls (QV33). PBHoover also detected antibiotic-resistant subpopulations that went undetected by Sanger. Using C1 chemistry, subpopulations as small as 9% of the total colony can be detected by PBHoover. This provides the most sensitive amplification-free molecular method for heterogeneity analysis and is in line with phenotypic methods’ sensitivity. This sensitivity significantly improves with the greater depth and lower error rate of the newer chemistries.Availability and ImplementationExecutables are freely available under GNU GPL v3+ athttp://www.gitlab.com/LPCDRP/pbhooverandhttp://www.gitlab.com/LPCDRP/CigarRoller. PBHoover is also available on bioconda:https://anaconda.org/bioconda/pbhoover.Contactfaramarz@sdsu.edu