Table 1 Comparison of different tools for the analysis of MAVE data. In our evaluations, we compare mutscan to DiMSum and Enrich2, as these are widely used in the field, and align with mutscan in terms of their scope and aim
Aspect | mutscan | DiMSum [20] | Enrich2 [19] |
|---|---|---|---|
OS availability | All | Unix-like | All (requires Python 2.7, end of life since January 1, 2020) |
External dependencies (in addition to R/Python packages) | None | None | |
Diagnostic report | Yes | Yes | Yes (autogenerated plots) |
Parallelizable | Yes | Yes | No |
Supported library design | Single-end, paired-end | Single-end, paired-end | Single-end, paired-end |
Statistical test specification | Any fixed effect model (formula) | No test. Calculates an enrichment score and error, limited to 9 replicates | One score per replicate, combined using a mixed-effects model |
Modularity | Samples are processed independently. Modularity between workflow steps | The entire experiment is processed together. Modularity between workflow steps if intermediate files are retained | The entire experiment is processed together. Intermediate results are reused if present |
Installation package | R package | Bioconda package | None |
Enrichment scores | Absolute and relative | Relative | Absolute and relative |
Use of barcodes or variant sequence | Both | Both | Both |
Alignment strategy | Tree-based string matching (Hamming or Levenshtein distance) | Clustering (Levenshtein distance, using starcode [24]) | String matching or gapped alignment (Needleman-Wunsch) |
UMIs | Yes | No | No |
Filtering options | Read length consistency, forbidden codons, base quality, nbr mismatches in constant sequence, nbr mutated bases/codons, nbr N bases | Minimal read length, base quality, expected errors, nbr mutated bases/amino acids, min nbr reads, invalid overlap, forbidden substitution, mutation type | Base quality, min nbr reads, nbr mutations, nbr N bases, invalid overlap |
Multiple amplicons | Yes | No | No |