Computational biologists have introduced DeepMutSim, an open-source Python tool designed to systematically generate every theoretically possible single-nucleotide variant and amino-acid substitution across human genes. Reported by researchers Liu Sun, Rongshu Wang, and Jian Yang, the software establishes an automated framework to perform comprehensive in silico variant enumeration. The system operates directly on canonical reference transcripts to assemble standardized variant libraries required for downstream genomic applications.
The generation of exhaustive variant catalogs addresses a persistent preparatory challenge in genomic medicine and experimental biology. While the accurate interpretation and annotation of genetic changes remain essential to the field, researchers constructing deep mutational scanning experiments, designing functional assays, or training machine-learning models frequently require complete variant reference sets for target genes. Dedicated software for this specific preparatory step has historically been lacking, leaving investigators without a standardized method to simulate all theoretical substitutions.
To establish a uniform baseline, DeepMutSim takes a human gene symbol and retrieves its corresponding MANE Select transcript, with optional support for MANE Plus Clinical records, as the canonical sequence reference. The program systematically enumerates all three alternative nucleotides across multiple sequence compartments. These include the amino-acid-encoding portion of the coding sequence, the 5-prime untranslated region, the 3-prime untranslated region, the canonical splice-site dinucleotides, and extended intronic splice-site regions.
The platform also maps functional protein variations by calculating all 19 possible alternative amino-acid substitutions for every residue and all 63 alternative codons for each codon position. By design, the terminal stop codon of the coding sequence is excluded from enumeration. To preserve sequence integrity at non-coding boundaries, reference bases at intronic splice-site positions are retrieved directly from official RefSeqGene records rather than inferred or assumed.
Every variant identified by the pipeline is cataloged using official Human Genome Variation Society nomenclature and mapped to corresponding GRCh38 genomic coordinates. In an initial demonstration on the human G6PD MANE Select transcript (NM_001360016.2), the software cataloged 4,635 coding sequence single-nucleotide variants, 198 variants in the 5-prime untranslated region, and 1,827 variants in the 3-prime untranslated region. The analysis also identified 132 canonical splice-site variants and 1,122 variants across extended splice-site regions.
At the protein and codon levels, the evaluation of G6PD produced 9,785 unique amino-acid substitutions alongside 32,445 total codon substitutions, of which 29,668 represented missense changes. To verify coordinate precision and sequence handling, the researchers subjected the software to an offline test suite comprising 80 unit checks. The team further validated approximately 400,000 generated genomic coordinates across five test genes—INS, TMPRSS6, ADAMTS13, SLC22A5, and CDKN2A—spanning both genomic strands against reference chromosome sequences, reporting zero coordinate mismatches.
The developers emphasize distinct functional boundaries for the utility of DeepMutSim. The software functions exclusively as an enumeration engine for predefined variant classes; it does not process raw sequencing reads to detect variants, nor does it generate functional predictions, pathogenicity scores, population allele frequencies, or clinical interpretations. By maintaining a lightweight architecture with minimal dependencies, the package integrates into bioinformatics workflows supporting experimental deep mutational scanning design and the characterization of variant landscapes in disease-associated genes.
Subject of Research: In silico enumeration of theoretical human gene single-nucleotide variants and amino-acid substitutions
Article Title: DeepMutSim: exhaustive in silico enumeration of human gene SNVs and amino-acid substitutions based on MANE Select transcripts in HGVS nomenclature
Article References: Sun, L., Wang, R., & Yang, J. (2026). DeepMutSim: exhaustive in silico enumeration of human gene SNVs and amino-acid substitutions based on MANE Select transcripts in HGVS nomenclature. BMC Genomics. https://doi.org/10.1186/s12864-026-13602-0
Image Credits: AI Generated
DOI: 10.1186/s12864-026-13602-0
Keywords: DeepMutSim, genomics, bioinformatics, variant enumeration, single-nucleotide variants, amino-acid substitutions, MANE Select, HGVS nomenclature, GRCh38, deep mutational scanning, computational biology, G6PD, splice-site variants, coding sequence, RefSeqGene, open-source software
News Source: Juliet Wilcox. (November 20, 2026). DeepMutSim Enumerates Theoretical Human Gene Variants Using Canonical Transcripts and HGVS Standards. Scienmag.



