Dataset Information

GLM-based optimization of NGS data analysis: A case study of Roche 454, Ion Torrent PGM and Illumina NextSeq sequencing data.

ABSTRACT:

Background

There are various next-generation sequencing techniques, all of them striving to replace Sanger sequencing as the gold standard. However, false positive calls of single nucleotide variants and especially indels are a widely known problem of basically all sequencing platforms.

Methods

We considered three common next-generation sequencers-Roche 454, Ion Torrent PGM and Illumina NextSeq-and applied standard as well as optimized variant calling pipelines. Optimization was achieved by combining information of 23 diverse parameters characterizing the reported variants and generating individually calibrated generalized linear models. Models were calibrated using amplicon-based targeted sequencing data (19 genes, 28,775 bp) from seven to 12 myelodysplastic syndrome patients. Evaluation of the optimized pipelines and platforms was performed using sequencing data from three additional myelodysplastic syndrome patients.

Results

Using standard analysis methods, true mutations were missed and the obtained results contained many artifacts-no matter which platform was considered. Analysis of the parameters characterizing the true and false positive calls revealed significant platform- and variant specific differences. Application of optimized variant calling pipelines considerably improved results. 76% of all false positive single nucleotide variants and 97% of all false positive indels could be filtered out. Positive predictive values could be increased by factors of 1.07 to 1.27 in case of single nucleotide variant calling and by factors of 3.33 to 53.87 in case of indel calling. Application of the optimized variant calling pipelines leads to comparable results for all next-generation sequencing platforms analyzed. However, regarding clinical diagnostics it needs to be considered that even the optimized results still contained false positive as well as false negative calls.

SUBMITTER: Sandmann S

PROVIDER: S-EPMC5319672 | biostudies-literature | 2017

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

GLM-based optimization of NGS data analysis: A case study of Roche 454, Ion Torrent PGM and Illumina NextSeq sequencing data.

Sandmann Sarah S de Graaf Aniek O AO van der Reijden Bert A BA Jansen Joop H JH Dugas Martin M

PloS one 20170221 2

<h4>Background</h4>There are various next-generation sequencing techniques, all of them striving to replace Sanger sequencing as the gold standard. However, false positive calls of single nucleotide variants and especially indels are a widely known problem of basically all sequencing platforms.<h4>Methods</h4>We considered three common next-generation sequencers-Roche 454, Ion Torrent PGM and Illumina NextSeq-and applied standard as well as optimized variant calling pipelines. Optimization was a ...[more]

PMID: 28222155

Similar Datasets

Project description:The emergence of benchtop sequencers has made clinical genetic testing using next-generation sequencing more feasible. Ion Torrent's PGM™ is one such benchtop sequencer that shows clinical promise in detecting single nucleotide variations (SNVs) and microindel variations (indels). However, the large number of false positive indels caused by the high frequency of homopolymer sequencing errors has impeded PGM™'s usage for clinical genetic testing. An extensive analysis of PGM™ data from the sequencing reads of the well-characterized genome of the Escherichia coli DH10B strain and sequences of the BRCA1 and BRCA2 genes from six germline samples was done. Three commonly used variant detection tools, SAMtools, Dindel, and GATK's Unified Genotyper, all had substantial false positive rates for indels. By incorporating filters on two major measures we could dramatically improve false positive rates without sacrificing sensitivity. The two measures were: B-Allele Frequency (BAF) and VARiation of the Width of gaps and inserts (VARW) per indel position. A BAF threshold applied to indels detected by UnifiedGenotyper removed ~99% of the indel errors detected in both the DH10B and BRCA sequences. The optimum BAF threshold for BRCA sequences was determined by requiring 100% detection sensitivity and minimum false discovery rate, using variants detected from Sanger sequencing as reference. This resulted in 15 indel errors remaining, of which 7 indel errors were removed by selecting a VARW threshold of zero. VARW specific errors increased in frequency with higher read depth in the BRCA datasets, suggesting that homopolymer-associated indel errors cannot be reduced by increasing the depth of coverage. Thus, using a VARW threshold is likely to be important in reducing indel errors from data with higher coverage. In conclusion, BAF and VARW thresholds provide simple and effective filtering criteria that can improve the specificity of indel detection in PGM™ data without compromising sensitivity.

Project description:BackgroundMassively parallel sequencing (MPS) technologies have the capacity to sequence targeted regions or whole genomes of multiple nucleic acid samples with high coverage by sequencing millions of DNA fragments simultaneously. Compared with Sanger sequencing, MPS also can reduce labor and cost on a per nucleotide basis and indeed on a per sample basis. In this study, whole genomes of human mitochondria (mtGenome) were sequenced on the Personal Genome Machine (PGMTM) (Life Technologies, San Francisco, CA), the out data were assessed, and the results were compared with data previously generated on the MiSeqTM (Illumina, San Diego, CA). The objectives of this paper were to determine the feasibility, accuracy, and reliability of sequence data obtained from the PGM.Results24 samples were multiplexed (in groups of six) and sequenced on the at least 10 megabase throughput 314 chip. The depth of coverage pattern was similar among all 24 samples; however the coverage across the genome varied. For strand bias, the average ratio of coverage between the forward and reverse strands at each nucleotide position indicated that two-thirds of the positions of the genome had ratios that were greater than 0.5. A few sites had more extreme strand bias. Another observation was that 156 positions had a false deletion rate greater than 0.15 in one or more individuals. There were 31-98 (SNP) mtGenome variants observed per sample for the 24 samples analyzed. The total 1237 (SNP) variants were concordant between the results from the PGM and MiSeq. The quality scores for haplogroup assignment for all 24 samples ranged between 88.8%-100%.ConclusionsIn this study, mtDNA sequence data generated from the PGM were analyzed and the output evaluated. Depth of coverage variation and strand bias were identified but generally were infrequent and did not impact reliability of variant calls. Multiplexing of samples was demonstrated which can improve throughput and reduce cost per sample analyzed. Overall, the results of this study, based on orthogonal concordance testing and phylogenetic scrutiny, supported that whole mtGenome sequence data with high accuracy can be obtained using the PGM platform.

Dataset Information

GLM-based optimization of NGS data analysis: A case study of Roche 454, Ion Torrent PGM and Illumina NextSeq sequencing data.

Background

Methods

Results

Publications

GLM-based optimization of NGS data analysis: A case study of Roche 454, Ion Torrent PGM and Illumina NextSeq sequencing data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets