m3triq 0.2.12 → 0.2.14

package/README.md

@@ -85,7 +85,7 @@ m3t md results <job-id>
 
 ```bash
 m3t predict esmfold MKFLILLFNILCL...        # Fast (~10s, max 1024aa)
-m3t predict alphafold2 MKFLILLFNILCL...     # Accurate (~15-20min, max 2048aa)
+m3t predict alphafold2 MKFLILLFNILCL...     # AF2-ptm monomer via deep MSA + ColabFold (~5-15min)
 
 # ESMFold2-Fast — monomer or multi-chain complex (self-hosted A100, max 2048aa total).
 # Beats AlphaFold3 on antibody-antigen DockQ from single sequence; returns pLDDT/pTM/ipTM.
@@ -95,8 +95,31 @@ m3t predict esmfold2-batch inputs.json                        # fold N complexes
 
 # Boltz-2 — biomolecular complex (protein + DNA/RNA + ligand) with binding affinity
 m3t predict boltz2 --protein MKFL... --ligand "CC(=O)O" --rna GGUC...
+
+# OpenFold3 — AlphaFold3-class complex (protein + DNA/RNA + ligand); real MSAs
+m3t predict openfold3 --protein EVQL... --protein DIQM...      # complex, auto-paired
+m3t predict openfold3 --protein MKFL... --ligand ATP --depth exhaustive
 ```
 
+## MSA Generation
+
+Generate a multiple-sequence alignment (`.a3m`) — the evolutionary input folders use.
+Two orthogonal knobs: **`--depth`** = coverage (databases), **`--pair`** = species
+pairing (only matters for multi-chain complexes; on by default, a monomer is never paired).
+
+```bash
+m3t msa --sequence MKFL...                                   # standard, monomer
+m3t msa --chain A:EVQL... --chain B:DIQM...                  # standard complex, auto-paired
+m3t msa --chain A:EVQL... --chain B:DIQM... --no-pair        # complex, force unpaired
+m3t msa --sequence MKFL... --depth exhaustive --templates    # + metagenomic envDB + PDB templates
+m3t msa --sequence MKFL... --depth custom \
+        --databases uniref30,envdb --max-sequences 300       # fine control (max_sequences caps depth)
+```
+
+- `--depth`: `standard` (UniRef30, default) | `exhaustive` (+ metagenomic envDB) | `custom`
+- `--templates`: also save per-chain PDB structural templates (`templates_<chain>.json`)
+- Output: one `.a3m` per chain in the project's Files tab; repeat sequences are cached (instant)
+
 ## Protein Embeddings
 
 ```bash

package/dist/cli.js

@@ -25,7 +25,7 @@ const program = new Command();
 program
     .name('m3t')
     .description('M3TRIQ — protein-ligand analysis from the terminal')
-    .version('0.2.12')
+    .version('0.2.14')
     .option('--json', 'Output as JSON (machine-readable)')
     .hook('preAction', (thisCommand) => {
     const opts = thisCommand.optsWithGlobals();

package/dist/client.d.ts

@@ -87,7 +87,7 @@ export declare class M3triqClient {
             id: string;
             sequence: string;
         }[];
-        depth: 'fast' | 'deep' | 'exhaustive' | 'custom';
+        depth: 'standard' | 'exhaustive' | 'custom' | 'fast' | 'deep';
         pair: boolean;
         templates?: boolean;
         custom?: {

package/dist/commands/msa.js

@@ -4,32 +4,34 @@ import { requireProject } from '../config.js';
 import { output } from '../output.js';
 import { jobUrl, maybeOpenBrowser } from '../url.js';
 // m3t msa — generate a multiple-sequence alignment (a3m) for one or more chains,
-// all on the self-hosted ColabFold MSA-search engine (GPU-MMseqs2):
-//   fast        = UniRef30, shallow, no pairing (quick triage)
-//   deep        = UniRef30, full depth + species pairing (complexes/hard families)
-//   exhaustive  = deep + the metagenomic envDB (remote-homology / orphan families)
-//   custom      = you pick the databases (+ max_sequences / pairing)
+// all on the self-hosted ColabFold MSA-search engine (GPU-MMseqs2). Two independent
+// knobs:
+//   --depth  = COVERAGE (which databases): standard (UniRef30) | exhaustive (+envDB,
+//              for remote-homology/orphan families) | custom (pick databases)
+//   --pair   = species PAIRING, only meaningful for a multi-chain complex (on by
+//              default; a monomer is never paired). --no-pair to force unpaired.
 // --templates also searches the PDB and saves per-chain structural templates.
 // Produces a downloadable .a3m artifact per chain (+ templates_<chain>.json).
-const MSA_DEPTHS = ['fast', 'deep', 'exhaustive', 'custom'];
+// (Legacy --depth fast/deep still accepted: fast = standard+unpaired, deep = standard.)
+const MSA_DEPTHS = ['standard', 'exhaustive', 'custom', 'fast', 'deep'];
 export function registerMsaCommands(program) {
     program.command('msa')
         .description('Generate a multiple-sequence alignment (a3m) — per-chain and (for complexes) species-paired')
         .option('--chain <id:seq>', 'Chain as "id:sequence" or "id:@path". Repeatable for multi-chain (paired) MSAs.', collectArg, [])
         .option('--sequence <seq>', 'Single-chain shortcut (chain id "A"). Use --chain for multi-chain.')
-        .option('--depth <tier>', 'MSA depth: fast | deep | exhaustive | custom (see `m3t msa --help`)', 'fast')
+        .option('--depth <coverage>', 'Coverage / databases: standard (UniRef30) | exhaustive (+ envDB) | custom', 'standard')
         .option('--templates', 'Also search the PDB for structural templates; saves a templates_<chain>.json per chain.')
         .option('--databases <csv>', 'For --depth custom: comma-separated database ids (e.g. "uniref30,envdb"). See get_msa_capabilities.')
         .option('--max-sequences <n>', 'For --depth custom: alignment depth/breadth cap.', (v) => parseInt(v, 10))
-        .option('--pair', 'Compute species-paired MSA across chains (complexes). Default on for ≥2 distinct chains.')
-        .option('--no-pair', 'Skip paired MSA (per-chain only).')
+        .option('--pair', 'Species-paired MSA — only matters for a multi-chain complex (on by default; a monomer is never paired).')
+        .option('--no-pair', 'Force unpaired (per-chain only), even for a complex.')
         .option('--name <name>', 'Custom job title')
         .action(async (opts) => {
         await runMsa(opts);
     });
 }
 async function runMsa(opts) {
-    const depth = (opts.depth || 'fast').toLowerCase();
+    const depth = (opts.depth || 'standard').toLowerCase();
     if (!MSA_DEPTHS.includes(depth)) {
         process.stderr.write(`Error: --depth must be one of ${MSA_DEPTHS.join(' | ')}, got "${opts.depth}".\n`);
         process.exit(1);
@@ -85,11 +87,9 @@ async function runMsa(opts) {
     const url = jobUrl(consoleUrl, project.id, result.job_id);
     maybeOpenBrowser(url);
     const timeEst = depth === 'exhaustive'
-        ? '~4-6 min (ColabFold UniRef30 + envDB on 2 T4s; +cold start if scaled to zero)'
-        : depth === 'fast'
-            ? '~2-4 min (ColabFold UniRef30; +cold start if scaled to zero)'
-            : '~3-5 min (ColabFold UniRef30 + paired; +cold start if scaled to zero)';
-    const extras = [pair ? 'paired' : '', templates ? '+templates' : ''].filter(Boolean).join(', ');
+        ? `~4-6 min (UniRef30 + envDB on 2 T4s${pair ? ' + paired' : ''}; +cold start if scaled to zero)`
+        : `~3-5 min (UniRef30${pair ? ' + paired' : ''}; +cold start if scaled to zero)`;
+    const extras = [pair ? 'paired' : 'unpaired', templates ? '+templates' : ''].filter(Boolean).join(', ');
     const lines = [
         `MSA job queued (${chains.length} chain${chains.length > 1 ? 's' : ''}, depth=${depth}${extras ? `, ${extras}` : ''})`,
         `Job ID: ${result.job_id.substring(0, 8)}`,

package/dist/commands/predict.js

@@ -13,19 +13,14 @@ export function registerPredictCommands(program) {
         .action(async (sequence, opts) => {
         await runPredict(sequence, 'esmfold', opts.name);
     });
-    // m3t predict alphafold2 — RETIRED (the alphafold2-nim VM was decommissioned in
-    // favour of ESMFold2). Kept as a signpost so scripts get a clear redirect instead
-    // of a job that silently fails on the deleted backend.
+    // m3t predict alphafold2 — AF2 (ptm) monomer via the self-hosted ColabFold service
+    // (the alphafold2-nim VM was retired; same AF2 weights, deep MSA from m3t_msa_server).
     predict.command('alphafold2')
-        .argument('[sequence]', 'Amino acid sequence (single letter codes)')
+        .argument('<sequence>', 'Amino acid sequence (single letter codes, max 2048aa)')
         .option('--name <name>', 'Name for the prediction')
-        .description('[RETIRED → use `predict esmfold2`] AlphaFold2 single-sequence prediction')
-        .action(async () => {
-        process.stderr.write('AlphaFold2 has been retired (the self-hosted backend was decommissioned).\n' +
-            'Use ESMFold2 instead — higher accuracy and it folds complexes too:\n' +
-            '  m3t predict esmfold2 --sequence <seq>            # monomer\n' +
-            '  m3t predict esmfold2 --chain A:<seq> --chain B:<seq>   # complex\n');
-        process.exit(2);
+        .description('AlphaFold2 (ptm) monomer prediction — deep MSA + ColabFold (~5-15 min, +cold start)')
+        .action(async (sequence, opts) => {
+        await runPredict(sequence, 'alphafold2', opts.name);
     });
     // m3t predict esmfold2 — monomer OR multi-chain (antibody-antigen, PPI)
     predict.command('esmfold2')
@@ -101,7 +96,7 @@ async function runPredict(sequence, method, name) {
     });
     const url = jobUrl(consoleUrl, project.id, result.job_id);
     maybeOpenBrowser(url);
-    const timeEst = method === 'alphafold2' ? '~15-20 minutes (varies with length)' : '~10 seconds';
+    const timeEst = method === 'alphafold2' ? '~5-15 minutes (deep MSA + ColabFold; +cold start)' : '~10 seconds';
     const data = { job_id: result.job_id, method, sequence_length: sequence.length, url };
     output(data, `${method} prediction created\nJob ID: ${result.job_id.substring(0, 8)}\nLength: ${sequence.length} residues\nEstimated: ${timeEst}`);
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "m3triq",
-  "version": "0.2.12",
+  "version": "0.2.14",
   "description": "M3TRIQ \u2014 protein-ligand analysis from the terminal",
   "type": "module",
   "bin": {