Platforms: osx-arm64, linux-aarch64; tests workflow on Linux and macOS; arm64 image

.github/workflows/docker.yml

@@ -42,6 +42,8 @@ jobs:
             type=ref,event=pr
             type=sha,format=short
 
+      - uses: docker/setup-qemu-action@v3
+
       - uses: docker/setup-buildx-action@v3
 
       - name: Build (load locally for the smoke test)
@@ -81,6 +83,7 @@ jobs:
         with:
           context: .
           push: true
+          platforms: linux/amd64,linux/arm64
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           build-args: VERSION=${{ steps.meta.outputs.version }}

.github/workflows/tests.yml

@@ -0,0 +1,32 @@
+name: tests
+
+# The test suite in the pixi `test` environment on Linux x86-64 and on macOS (Apple Silicon).
+on:
+  pull_request:
+  push:
+    branches: [main]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  pytest:
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, macos-latest]
+    runs-on: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v4
+      - uses: prefix-dev/[email protected]
+        with:
+          environments: test
+          locked: true
+      - name: Tests
+        run: pixi run -e test pytest tests -q
+      - name: Smoke test
+        run: |
+          pixi run matpredict --version
+          pixi run matpredict detect --help > /dev/null
+          pixi run bash -c 'for t in tblastn diamond miniprot exonerate mafft augustus taxonkit; do command -v $t > /dev/null || { echo "missing $t"; exit 1; }; done'

CHANGELOG.md

@@ -7,6 +7,7 @@ release completed.
 ## [Unreleased]
 
 ### Added
+- Platforms: `pixi.toml` and `pixi.lock` now cover `osx-arm64` and `linux-aarch64` as well as `linux-64`; a `tests` workflow runs the suite (`pixi run -e test pytest tests`) and a smoke test on Linux and macOS; the Docker image is built for `linux/amd64` and `linux/arm64`. The re-lock also moved the `linux-64` pins, so run the regression check against it.
 - Five accepted tier-3 records (curator sign-off 2026-10-30, J. Stajich): *Saccharomycopsis schoenii* GDD12 MATss (MATa-MATalpha left cluster and MATa right cluster, `db/Ascomycota/Ascoideales/`), *Barnettozyma californica* PB4207 MATbz and *Petasospora fabianii* MATpf (JOY008 MATa, TBRC 4498 MATalpha, `db/Ascomycota/Phaffomycetales/`). The gene models are our own (RNA-seq for *S. schoenii*, sequence-based for the others). With them, 20 of the 21 genomes that lose every MATsc/MTL call under the v0.8 cross-family rules are called again by their own family (trial run, `--preset v0.8`); *Ogataea degrootiae* GCA_019232045.1 stays uncalled.
 - Raw-reads wave 2 (`datasets/reads/sra_survey/pilot/wave2/`): 55 more tier A, A1 and B species typed (1,468 runs; 13 confirmed, 16 provisional, 7 provisional-homothallic, 49 withheld), yeast idiomorph names (a / alpha, P / M, A / B) carried through, *Botrytis cinerea* typed on all 333 eligible samples. Species reports gain an allele-count table for all studies pooled or study by study, per run or per clone, and clone groups for every species with 10 or more sketched runs (average linkage at 0.998; counts per clone only where the clone number is stable). Output only; the genome tables and the database are unchanged.
 - Idiomorph labels beside the raw names (curator rulings of 2026-10-28): `scripts/idiomorph_label_registry.json` (lineage and reference-organism codes, cell tokens, gene maps), `scripts/idiomorph_labels.py`, `scripts/label_loci_table.py`; the site data carry `label_alleles` in `taxa.json`, `call_label` / `alleles_label` in the genome tables and `data/idiomorph_labels.json`, and the explorer shows `Lineage__System_Cell` labels (for example `Taphr__SchPom_M`, `Sacch__CanAlb_alpha`, `Peziz__AspFum_MAT1-1`) with the raw name and the reference organism in the tooltip, and a switch back to the raw names. Output only: the database, the reports and the raw columns are unchanged. All 27,885 v0.7.0 loci are labelled. Duplicate family calls (`scripts/dedupe_family_calls.py`) are not removed from the counts.

README.md

@@ -196,6 +196,8 @@ apptainer pull matpredict.sif docker://ghcr.io/stajichlab/matpredict:latest
 apptainer run matpredict.sif detect --genome genome.fna --taxid 4837 --out-dir out
 ```
 
+**macOS (Apple Silicon).** `pixi install` works natively on `osx-arm64` (every tool in `pixi.toml` resolves there); the Docker image is also published for `linux/arm64`.
+
 **Conda / mamba.** [environment.yml](environment.yml) lists the same packages
 as `pixi.toml` (exported with `pixi workspace export conda-environment`). Run
 from the repository root, because it installs MATPredict from the checkout: