Excerno: Using Mutational Signatures in Sequencing Data to Filter False Variants Caused by Clinical Archival
Authors:
- 1 April 2023
- journal article
- research article
- Published by SAGE Publications in Journal of Computational Biology
- Vol. 30 (4) , 366-375
Abstract
The accurate detection of point mutations from pathology slides using sequencing data is of great importance in cancer genomics and precision oncology. Formalin-fixation paraffin-embedding (FFPE) is a widely used technique to preserve pathology tissues. The FFPE process introduces artificial C > T mutations in next-generation sequencing, so we set out to develop excerno , a method to score and filter such spurious variants. FFPE mutational artifacts follow a mutational signature. By using the FFPE signature and Bayes' formula, we can calculate the probability of a mutation resulting from the FFPE process and use this probability to filter FFPE variants. We implement this method as the excerno R package. We tested excerno by simulating mutations across all 60-baseline mutational signatures from the Catalog of Somatic Mutations in Cancer (COSMIC) and combining them with mutations following the FFPE mutational signature. The sensitivity and specificity of excerno are adversely affected by the cosine similarity between the baseline and FFPE signatures ( c o s F F P E ). Higher percentages of FFPE mutations ( p c t F F P E ) result in increased sensitivity and reduced specificity. The specificity and sensitivity of excerno can be predicted as linear model with an interaction term using c o s F F P E and p c t F F P E , with an R 2 o f 0 . 8 4 and 0.79, respectively. Finally, we tested excerno using six RNA sequencing cancer samples and observed concordant trends of specificity and sensitivity with respect to our simulated data. The excerno R package can be used to annotate and filter FFPE-induced mutations in cancer genomics. Our method is adversely affected by c o s F F P E and p c t F F P E .Keywords
This publication has 20 references indexed in Scilit:
- MutationalPatterns: comprehensive genome-wide analysis of mutational processesGenome Medicine, 2018
- Mutational signatures associated with tobacco smoking in human cancerScience, 2016
- Exonuclease mutations in DNA polymerase epsilon reveal replication strand specific mutation patterns and human origins of replicationGenome Research, 2014
- Cytosine Deamination Is a Major Cause of Baseline Noise in Next-Generation SequencingMolecular Diagnosis & Therapy, 2014
- Signatures of mutational processes in human cancerNature, 2013
- Deciphering Signatures of Mutational Processes Operative in Human CancerCell Reports, 2013
- Towards accurate detection and genotyping of expressed variants from whole transcriptome sequencing dataBMC Genomics, 2012
- SomaticSniper: identification of somatic point mutations in whole genome sequencing dataBioinformatics, 2011
- Mapping short DNA sequencing reads and calling variants using mapping quality scoresGenome Research, 2008
- Learning the parts of objects by non-negative matrix factorizationNature, 1999