swiftomics About

What is Swiftomics, and how does it work?

Developed by Amulya Baral

Swiftomics is an open-source, privacy-friendly, zero-install pathogen profiling platform.

It uses KMA (read-based gene detection), version 1.5.1, compiled to WebAssembly.

All of your data is analyzed locally in your own browser, and does not have to be uploaded to a server anywhere. This makes it fast, private, and safe to use even with sensitive clinical or unpublished sequencing data.

Because everything runs client-side, there is no account to create and no server that you have to put your trust on. The only network traffic is the one-time download of the WebAssembly engine and whichever reference database you select.

Databases are downloaded once and cached in your browser (IndexedDB), so repeat runs against the same database start instantly. The first run against a large database may take a moment while it downloads.

Getting started

  1. Pick your input type — paired-end, single-end, or Nanopore.
  2. Choose the reference database you want to align the reads to.
  3. Upload your read file(s) (.fastq or gzipped .fastq.gz), or click Try example data to run a bundled example sample.
  4. Hit Run analysis. Live progress is streamed into the Logs panel that you can expand any time to watch the KMA command and its output.

Databases for read-based gene detection

Right now, we support these databases:

ResFinder 2.6.0
Acquired AMR genes (CGE).
CARD 4.0.1
Comprehensive Antibiotic Resistance Database, protein homolog model.
VFDB
Virulence Factor Database Set A (core).
NCBI 16S rRNA
Broad species identification via KMA -Sparse (RefSeq Targeted Loci, 20,292 species). 16S resolves genus reliably; closely related species may be indistinguishable.
Clinical genome panel
Whole-genome species identification for 104 clinical species. This typically resolves what 16S cannot (e.g. Escherichia vs Shigella). It is a closed set and organisms outside it map to the nearest member.
MLST (PubMLST)
Classic multi-locus sequence typing for 175 organisms. Reads are aligned to every known allele of each housekeeping locus; the best allele per locus gives an allelic profile, which is looked up to report the sequence type (ST). Data is the freely-redistributable pre-2025 PubMLST snapshot (alleles/profiles submitted on or before 2024⁠-⁠12⁠-⁠31). A few organisms whose schemes are hosted elsewhere (notably Klebsiella pneumoniae and Listeria monocytogenes, at the Institut Pasteur) are not yet included.

Some notes

  • Defaults of 80% identity and 80% coverage have been chosen following common reporting thresholds; this does not in any way mean they are absolute. For example, you can lower the identity threshold to recover divergent variants, or lower the coverage threshold for fragmented or low-depth samples.
  • The identity and coverage sliders apply to the gene-detection databases (ResFinder, CARD, VFDB). The species-identification databases (16S and the clinical genome panel) use KMA's -Sparse k-mer counting instead of alignment, so those sliders are hidden for them.
  • For 16S, calls that match several species equally well are reported at the genus or group level rather than guessing a single species. The clinical genome panel can often resolve them, given it is on the species list.
  • For MLST, choose the organism first, then run. A sequence type is reported when the full allele profile matches a known ST; otherwise the result is a novel ST (new combination of known alleles), a novel/uncertain allele (a locus that didn't match well — possibly a new allele or low coverage), or an incomplete profile — with the nearest known ST shown. Per-locus identity, coverage, and depth indicate confidence; a locus a few percent under 100% is usually sequencing error, not a new allele.
  • Nanopore mode enables KMA's -bcNano for indel-aware base calling.
  • You can click any detected gene in the results to inspect its coverage, consensus sequence, alignment, and variants.
  • The output bundle includes .res, .aln, .fsa, .frag.gz, .mat.gz, .vcf.gz, and .mapstat.
  • Swiftomics needs a modern browser with SharedArrayBuffer support (recent Chrome, Firefox, Edge, or Safari).

Citation

We have a short communication in preparation. Please cite KMA and the relevant database in the meantime:

Clausen PTLC, Aarestrup FM, Lund O. Rapid and precise alignment of raw reads against redundant databases with KMA. BMC Bioinformatics. 2018;19:307. doi:10.1186/s12859-018-2336-6

For MLST results, please also cite PubMLST and the relevant scheme: Jolley KA, Bray JE, Maiden MCJ. Open-access bacterial population genomics: BIGSdb software, the PubMLST.org website and their applications. Wellcome Open Res. 2018;3:124. doi:10.12688/wellcomeopenres.14826.1