With the growing value of next generation sequencing (NGS) assays for the determination of minimal residual disease (MRD) in the clinic, the conﬁdent and sensitive detection of low frequency variants is crucial to the treatment of cancer. Current in silico pipelines often lack the sensitivity to detect low frequency variants, whose variant allele frequencies (VAFs) covary with sample purity (i.e. tumor-normal and/or normal-tumor contamination), sample clonality, and copy number variations. Sensitivity is also confounded by the background error inherent to sequencing data, which may be introduced by systematic platform error, library ampliﬁcation, and errors in sample preparation. Attempting to mitigate background error in sequencing data, researchers have developed many software error correction programs that model sources of error to mitigate its impact on downstream processing. While these models have been developed for de novo assembly, metagenomics research, and viral haplotype reconstruction, their application to the use case of low frequency variant detection has yet to be explored in-depth. For this research, we sought to develop a software framework for the evaluation of background error models in the low frequency variant use case, with a speciﬁc focus on their potential value to MRD monitoring.