Microphaser - a small-scale phasing approach for improved personalized neopeptidome creation

Motivation: Haplotype phasing approaches have been shown to improve accuracy of the search space of neoantigen prediction by determining if adjacent variants are located on the same chromosomal copy. However, the aneuploid nature of cancer cells as well as the admixture of different clones in tumor bulk sequencing data are challenging the current diploid based phasing algorithms. We present microphaser, a small-scale phasing approach to improve haplotyping variants in cancer samples. Microphaser aims to create a more accurate neopeptidome for downstream neoantigen prediction. Results: Microphaser achieves large concordance with state-of-the-art phasing-aware neoantigen prediction pipeline neoepiscope, with differences in edge cases and an improved filtering step. Availability: Microphaser is written in the Rust programming language. It is made available via Github (https://github.com/koesterlab/microphaser)and Bioconda. The corresponding prediction pipeline (https://github.com/snakemake-workflows/dna-seq-neoantigen-prediction) has been written within the Snakemake workflow management system and can be deployed as part of the snakemake-workflows project.