Bioinformatics

A VS Code extension for browsing protein multiple sequence alignments and HHsuite hit-list results without leaving the editor.

Features

MSA viewer (.a3m, .a2m, .fasta, .sto, .clu, .psi)

• Per-residue coloring with four palettes — Chemical (default), Clustal X, Taylor, and Zappo — switchable from the toolbar.
• Row virtualization: only rows in the viewport are in the DOM, so 3,000+ row alignments stay responsive.
• Sticky header stack: column ruler · coverage histogram · query row, all pinned to the top while you scroll.
• Coverage histogram, conservation-tinted — bar height = coverage, bar color = per-column conservation. Inline ↻ button on the row label cycles off → entropy (1 − H/ln 20) → BLOSUM62 SP; current mode + formula shown in the button's tooltip.
• Zoomable residue cells (toolbar + / −, Ctrl/Cmd + scroll, click the size label to reset to 10 px). Controls bar stays at UI size; only the alignment grid scales.
• Drag-resize coverage row (bottom edge) and drag-resize the name column (or pull all the way left to hide).
• Inserts toggle for a3m/a2m (show/collapse insert columns). Stockholm files automatically use #=GC RF to identify match vs. insert columns.
• Filter by name (Cmd/Ctrl + F) and goto column (Cmd/Ctrl + L) inputs.
• Per-column hover — column index, query residue, coverage, Shannon entropy (with per-column N_eff), BLOSUM62 sum-of-pairs (raw + normalized), and the top 5 residues with counts and percentages.
• Stats pills: ungapped query length, depth, HHblits-style N_eff (exp(mean per-column Shannon entropy), gaps excluded, restricted to query-residue columns), log₁₀H̄ and log₂H̄ of mean Hamming distance to the query.
• Drag-and-drop import — drop any supported alignment file onto the viewer to swap content in place.

HHR pairwise viewer (.hhr)

For HHsearch / HHblits / HHpred result files.

• HitMap — SVG visualization of every hit's coverage along the query. Hits are packed into rows (non-overlapping share rows), color-coded by probability, labeled with the template name and description (clipped to the rect's right edge), and clickable to scroll to the alignment block.
• Sortable hit table — No / Hit / Prob / E-value / Score / Cols / Q range / T range. Click any header to sort.
• Per-hit alignment blocks — query / consensus / agree-string / template, plus secondary-structure tracks (Q ss_pred, T ss_pred, T ss_dssp) and confidence rows when present.
• Auto-wrap to window width — chunks reflow on resize; type a number in the wrap input to override.
• Color toggle — off by default for pairwise viewing (the agree-string already conveys conservation); flip on for full per-residue coloring.
• Bidirectional jump: click a row in the table → scroll to its alignment block; click No N. in an alignment header → scroll to the row in the table (with a brief flash).

Templates view (a3m with [subseq from] headers)

When a manifest of templates is detected (e.g. hmmsearch output), an extra Templates toggle appears in the toolbar. Switching it on:

• Replaces the residue grid with a structured table — # / Accession / Range / Length / Description.
• Shows a HitMap above the table built from each row's first/last non-gap residue (click a rect to scroll the matching row).

Header parsing & database links

Headers in MSA labels, hit-list rows, and template tables are parsed and linkified.

UniProt FASTA headers (per the UniProt spec) are recognized in their canonical shapes, including the optional HHsuite /start-end subseq suffix:

• UniProtKB: >sp|P12345|FOO_HUMAN Protein name OS=Organism OX=NCBI_taxID GN=Gene PE=1 SV=2 (also tr|…)
• UniRef: >UniRef90_A0A2L2TCV1 Cluster name n=N Tax=Taxon TaxID=ID RepID=Rep (or …/start-end)
• UniParc: >UPI0000000001 status=active

The structured fields (entry name · protein · organism + taxID · gene · cluster size · representative · range · …) are surfaced as a multi-line tooltip on the head link.

Auto-linked accessions (head text → canonical entry page):

A small AFDB chip renders next to UniProt-kind links and after UniRef rep accessions, taking you straight to alphafold.ebi.ac.uk/entry/{acc} for the structural alternative.

Source-view syntax highlighting

TextMate grammars ship for the FASTA family (.fasta / .fas / .fa / .afa / .afas / .afasta / .a2m / .a3m), Stockholm (.sto / .stockholm), and Clustal (.clu / .aln). The FASTA grammar distinguishes match (uppercase), insert (lowercase), gap (-), and gap-vs-insert (.) per the a2m/a3m casing convention; Stockholm gets structured #=GF / #=GS / #=GR / #=GC annotations.

Install

From the VS Code Marketplace:

code --install-extension vv137xyz.vscode-bioinformatics

Or search for Bioinformatics in the Extensions sidebar (publisher: vv137xyz).

To install a specific build, download vscode-bioinformatics-X.Y.Z.vsix from the Releases page and either run code --install-extension <file>.vsix, or use Extensions sidebar → … menu → Install from VSIX….

Usage

Open any supported file. The viewer is registered with priority: "option", so the regular text editor opens by default. Switch to the viewer either way:

• Click the MSA button in the editor title bar.
• Run Bioinformatics: Open in MSA Viewer from the Command Palette (Cmd/Ctrl + Shift + P).
• Right-click the file → Open With… → MSA Viewer.
• Pin the viewer as the default for an extension by adding to settings.json:

  "workbench.editorAssociations": {
    "*.hhr": "bioinformatics.msaViewer",
    "*.a3m": "bioinformatics.msaViewer"
  }
  

Keyboard shortcuts

Development

git clone https://github.com/vv137/vscode-bioinformatics
cd vscode-bioinformatics
npm install
npm run compile
code --extensionDevelopmentPath=$(pwd)

The repo layout:

src/                     # extension (TypeScript)
  parsers/               # fas / a2m / a3m / sto / clu / psi / hhr
  viewer/                # CustomTextEditorProvider + message protocol
media/                   # webview assets (pure DOM)
  viewer.module.js       # window.MsaViewer.create — no host deps
  viewer.adapter.js      # VS Code wiring (~30 lines)
  viewer.css             # shared
syntaxes/                # TextMate grammars (fasta, stockholm, clustal)
references/Toolkit-audit.md  # feature audit + plan reference

The renderer in media/viewer.module.js has no VS Code dependencies — state persists via a Storage-shaped object passed in by the host. The same module ships into plmMSA's public/ folder with a 5-line wrapper using localStorage.

Acknowledgments

• Renderer originally ported from plmMSA.
• File-format conventions follow hh-suite's reformat.pl.
• Hit-list / pairwise viewer design inspired by proteinevolution/Toolkit (HHblits / HHpred / HMMER result tabs).
• Example fixtures from aqlaboratory/openfold (examples/6KWC_1/).

License

MIT © 2026 Minsoo Kim