Benchmarking Somatic Variant Callers: A Tale Unfolding In The Synthetic Genomics Feature Space

A) The coverage barplots and B) The AF barplots and C) The Δ 𝐴𝐹 density plots for all TP variants. True Positive SNVs Δ 𝐴𝐹 = 𝐴𝐹Caller - 𝐴𝐹GT . SYNTH4BENCH II Benchmarking Somatic Variant Callers A Tale Unfolding In The Synthetic Genomics Feature Space Styliani-Christina Fragkouli*, Nikos Pechlivanis, Anastasia Anastasiadou, Georgios Karakatsoulis, Aspasia Orfanou, Panagoula Kollia, Andreas Agathangelidis and Fotis Psomopoulos* A) The coverage barplots and B) The AF barplots for all FP variants. C) The coverage barplots and D) The AF barplots for all FN variants. False Positive & False Negative SNVs Background: Identifying cancer-related genomic alterations with algorithms is vital, but evaluating their accuracy is challenging due to the lack of high-quality datasets. To address this issue, we implemented synth4bench, a framework, which calls the NEATv3.3 simulator to generate synthetic genomics data, calls somatic variant callers and then benchmarks their results. Methods: We conducted a comprehensive evaluation of Mutect2, Freebayes, VarDict, VarScan2, and LoFreq, comparing their results with our synthetic ground truth. Synth4bench helps to elucidate the fundamental principles of each variant caller by exposing them to data across the synthetic genomics data feature space while inspecting their response. Results: Using synthetic datasets as ground truth is effective for benchmarking tumor-only somatic variant calling algorithms. Synth4bench leverages this effort by providing a ground truth dataset thus improving precision particularly for low frequency variants. This method enables transparent and direct algorithm comparisons, while exploring their behavior. GA No 101172872. The JU receives support from the EU Horizon Europe and COCIR, EFPIA, EuropaBio, MedTech Europe, and Vaccines Europe and DNV Funded by the IHI & European Union National and Kapodistrian University of Athens Faculty of Sciences Department of Biology Section of Genetics & Biotechnology [email protected] sfragkoul scfragkoul Callers vs Ground Truth Variant Calling Downstream Analysis Preprocess Synth Data Generation Reference CERTH Code Data Supported with Travel Fellowship by