TotalReCaller: improved accuracy and performance via integrated alignment and base-calling

Menges, F., Narzisi, G., Mishra, B. (September 2011) TotalReCaller: improved accuracy and performance via integrated alignment and base-calling. Bioinformatics, 27 (17). pp. 2330-2337. ISSN 1367-4803

Abstract

Motivation: Currently, re-sequencing approaches use multiple modules serially to interpret raw sequencing data from next-generation sequencing platforms, while remaining oblivious to the genomic information until the final alignment step. Such approaches fail to exploit the full information from both raw sequencing data and the reference genome that can yield better quality sequence reads, SNP-calls, variant detection, as well as an alignment at the best possible location in the reference genome. Thus, there is a need for novel reference-guided bioinformatics algorithms for interpreting analog signals representing sequences of the bases ({A, C, G, T}), while simultaneously aligning possible sequence reads to a source reference genome whenever available. Results: Here, we propose a new base-calling algorithm, TotalReCaller, to achieve improved performance. A linear error model for the raw intensity data and Burrows-Wheeler transform (BWT) based alignment are combined utilizing a Bayesian score function, which is then globally optimized over all possible genomic locations using an efficient branch-and-bound approach. The algorithm has been implemented in soft- and hardware [field-programmable gate array (FPGA)] to achieve real-time performance. Empirical results on real high-throughput Illumina data were used to evaluate TotalReCaller's performance relative to its peers-Bustard, BayesCall, Ibis and Rolexa-based on several criteria, particularly those important in clinical and scientific applications. Namely, it was evaluated for (i) its base-calling speed and throughput, (ii) its read accuracy and (iii) its specificity and sensitivity in variant calling.

Item Type: Paper
Uncontrolled Keywords: short read alignment genome generation ultrafast search blast tool
Subjects: bioinformatics
bioinformatics > genomics and proteomics
Investigative techniques and equipment
Investigative techniques and equipment > assays
bioinformatics > genomics and proteomics > computers > computer software
Investigative techniques and equipment > whole exome sequencing
Investigative techniques and equipment > assays > whole exome sequencing
CSHL Authors:
Communities: CSHL labs > Wigler lab
Depositing User: Matt Covey
Date: 1 September 2011
Date Deposited: 06 Feb 2013 17:44
Last Modified: 06 Feb 2013 17:44
Related URLs:
URI: https://repository.cshl.edu/id/eprint/27177

Actions (login required)

Administrator's edit/view item Administrator's edit/view item