VarFreqAnalyzer: A Computational Framework for Rapid Internal Variant Frequency Estimation from Large Institutional VCF Repositories

With the advent of next generation sequencing (NGS) there are many institutional variant call format (VCF) repositories generated from routine clinical and research activities. Retrieval of variant evidence is still inefficient, especially for lab decision making, at the institution level. This paper introduces the computational framework VarFreqAnalyzer that is built using Python, provides indexing of variants, robust exact matching of alleles, identification of carriers based on reported data, estimation of Internal Variant Frequency (IVF), visualization and standardized reporting. Structured genomic evidence retrieval is supported by the framework in one graphical interface. A repository of 23,798 VCF files was analyzed that were indexed in an institutional repository. In 81 min, a network-accessed query processed 21,791 files and identified 11 carrier observations (a preliminary file based IVF of 0.0505%) with a rate of 269 files/min. The local-storage observation took about 15 min for the oncology category alone. When comparing the programs with BCFtools and VCFtools, a focus on the integration into the workflow is emphasized, but not outright performance. The discrepancy in the number of files processed (2007) and the unverified carrier counts as well as denominator assumptions and the lack of controlled memory and runtime benchmarks, limit interpretation. VarFreqAnalyzer establishes a framework for traceable institutional variant evidence retrieval, with the aim of genomic human-supervised decision support, which needs to be explored further in an independent fashion.