Limits...
A simple data-adaptive probabilistic variant calling model.

Hoffmann S, Stadler PF, Strimmer K - Algorithms Mol Biol (2015)

Bottom Line: It performs specifically well in cases with low allele frequencies.The application to next-generation sequencing data reveals stark differences of the score distributions indicating a strong influence of data specific sources of noise.The proposed model is specifically designed to adjust to these differences.

View Article: PubMed Central - PubMed

Affiliation: Junior Research Group Transcriptome Bioinformatics, University Leipzig, Härtelstraße 16-18, Leipzig, Germany ; Interdisciplinary Center for Bioinformatics and Bioinformatics Group, University Leipzig, Härtelstraße 16-18, Leipzig, Germany ; LIFE Research Center for Civilization Diseases, University Leipzig, Härtelstraße 16-18, Leipzig, Germany.

ABSTRACT

Background: Several sources of noise obfuscate the identification of single nucleotide variation (SNV) in next generation sequencing data. For instance, errors may be introduced during library construction and sequencing steps. In addition, the reference genome and the algorithms used for the alignment of the reads are further critical factors determining the efficacy of variant calling methods. It is crucial to account for these factors in individual sequencing experiments.

Results: We introduce a simple data-adaptive model for variant calling. This model automatically adjusts to specific factors such as alignment errors. To achieve this, several characteristics are sampled from sites with low mismatch rates, and these are used to estimate empirical log-likelihoods. The likelihoods are then combined to a score that typically gives rise to a mixture distribution. From this we determine a decision threshold to separate potentially variant sites from the noisy background.

Conclusions: In simulations we show that our simple model is competitive with frequently used much more complex SNV calling algorithms in terms of sensitivity and specificity. It performs specifically well in cases with low allele frequencies. The application to next-generation sequencing data reveals stark differences of the score distributions indicating a strong influence of data specific sources of noise. The proposed model is specifically designed to adjust to these differences.

No MeSH data available.


Statistical performance measures on simulated data sets. The data adaptive model implemented in haarz was compared to SAMtools and GATK in terms of recall and positive predictive value. SNV calling was performed on twelve different data sets varying in the content of the variant allele (20% and 50%) as well as the simulated coverage (10-200). In all of these scenarios the data adaptive model is at par with both alternative callers.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4363181&req=5

Fig4: Statistical performance measures on simulated data sets. The data adaptive model implemented in haarz was compared to SAMtools and GATK in terms of recall and positive predictive value. SNV calling was performed on twelve different data sets varying in the content of the variant allele (20% and 50%) as well as the simulated coverage (10-200). In all of these scenarios the data adaptive model is at par with both alternative callers.

Mentions: All of the tested programs show a good recall and positive predictive value in all 12 simulations. For low allele contents in conjunction with low coverages, however, SAMtools attains comparably low positive predictive values. Surprisingly, after reaching a maximum recall for the coverage of 100, the recall drops substantially for coverage 200. For the simulations with 50% allele content, all tools show high recalls and good positive predictive values. Again, SAMtools achieves only a comparably low positive predictive value for poorly covered SNVs (Figure 4). Except for the lowest coverage, all tools performed well on these data sets.Figure 4


A simple data-adaptive probabilistic variant calling model.

Hoffmann S, Stadler PF, Strimmer K - Algorithms Mol Biol (2015)

Statistical performance measures on simulated data sets. The data adaptive model implemented in haarz was compared to SAMtools and GATK in terms of recall and positive predictive value. SNV calling was performed on twelve different data sets varying in the content of the variant allele (20% and 50%) as well as the simulated coverage (10-200). In all of these scenarios the data adaptive model is at par with both alternative callers.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4363181&req=5

Fig4: Statistical performance measures on simulated data sets. The data adaptive model implemented in haarz was compared to SAMtools and GATK in terms of recall and positive predictive value. SNV calling was performed on twelve different data sets varying in the content of the variant allele (20% and 50%) as well as the simulated coverage (10-200). In all of these scenarios the data adaptive model is at par with both alternative callers.
Mentions: All of the tested programs show a good recall and positive predictive value in all 12 simulations. For low allele contents in conjunction with low coverages, however, SAMtools attains comparably low positive predictive values. Surprisingly, after reaching a maximum recall for the coverage of 100, the recall drops substantially for coverage 200. For the simulations with 50% allele content, all tools show high recalls and good positive predictive values. Again, SAMtools achieves only a comparably low positive predictive value for poorly covered SNVs (Figure 4). Except for the lowest coverage, all tools performed well on these data sets.Figure 4

Bottom Line: It performs specifically well in cases with low allele frequencies.The application to next-generation sequencing data reveals stark differences of the score distributions indicating a strong influence of data specific sources of noise.The proposed model is specifically designed to adjust to these differences.

View Article: PubMed Central - PubMed

Affiliation: Junior Research Group Transcriptome Bioinformatics, University Leipzig, Härtelstraße 16-18, Leipzig, Germany ; Interdisciplinary Center for Bioinformatics and Bioinformatics Group, University Leipzig, Härtelstraße 16-18, Leipzig, Germany ; LIFE Research Center for Civilization Diseases, University Leipzig, Härtelstraße 16-18, Leipzig, Germany.

ABSTRACT

Background: Several sources of noise obfuscate the identification of single nucleotide variation (SNV) in next generation sequencing data. For instance, errors may be introduced during library construction and sequencing steps. In addition, the reference genome and the algorithms used for the alignment of the reads are further critical factors determining the efficacy of variant calling methods. It is crucial to account for these factors in individual sequencing experiments.

Results: We introduce a simple data-adaptive model for variant calling. This model automatically adjusts to specific factors such as alignment errors. To achieve this, several characteristics are sampled from sites with low mismatch rates, and these are used to estimate empirical log-likelihoods. The likelihoods are then combined to a score that typically gives rise to a mixture distribution. From this we determine a decision threshold to separate potentially variant sites from the noisy background.

Conclusions: In simulations we show that our simple model is competitive with frequently used much more complex SNV calling algorithms in terms of sensitivity and specificity. It performs specifically well in cases with low allele frequencies. The application to next-generation sequencing data reveals stark differences of the score distributions indicating a strong influence of data specific sources of noise. The proposed model is specifically designed to adjust to these differences.

No MeSH data available.