A cancer cell-line titration series for evaluating somatic classification

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• Accurate detection of somatic variants in tumor-normal sequencing is challenging due to normal cell contamination and tumor heterogeneity.
• Sub-clonal variants at low allele frequencies are difficult to identify reliably.

Purpose of the Study:

• To present a cell-line titration series dataset for evaluating somatic variant calling pipelines.
• To enable reliable calling of true somatic mutations at low allele frequencies.

Main Methods:

• Generated samples by mixing cell-line DNA with matched normal DNA at 8 ratios, creating known tumor cellularities.
• Performed exome sequencing at >300× depth on Illumina HiSeq.
• Processed data with multiple variant calling pipelines and validated >1500 somatic variant candidates using Ion Torrent PGM.

Main Results:

• The best performing pipelines maintained high precision across all tested cellularities.
• Estimated the number of true somatic variants missed as cellularity and coverage decreased.

Conclusions:

• The cell-line titration dataset and verification results effectively evaluated somatic variant calling pipelines.
• This dataset will be valuable for future development of somatic calling algorithms.
• Data is publicly available via the European Genome-phenome Archive (EGAS00001001016).