Optional output alignments

Co-authored-by: aziele <a.zielezinski@gmail.com>
Co-authored-by: Sebastian Deorowicz <sebastian.deorowicz@polsl.pl>

.github/workflows/deploy.yml

@@ -0,0 +1,92 @@
+name: Deploy
+
+on:
+ release:
+    types:
+      - created
+      - updated
+
+jobs:
+
+
+ ########################################################################################
+  checkout:
+    name: Checkout
+    strategy:
+      matrix:
+        machine: [x64_linux, x64_mac, arm64_linux, arm64_mac]
+    runs-on: [self-hosted, vclust, '${{ matrix.machine }}']
+
+    steps:
+    - uses: actions/checkout@v4
+      with:
+        submodules: recursive
+    - name: Get tags
+      run: | 
+        cd ./3rd_party/clusty/libs/igraph
+        git fetch --prune --unshallow
+        echo exit code $?
+        git tag --list
+      continue-on-error: true
+
+########################################################################################
+  make:
+    name: Make
+    needs: checkout
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [x64_linux, arm64_linux]
+        leiden: [true]
+        include:
+        - machine: x64_mac
+          leiden: false
+        - machine: arm64_mac
+          leiden: false
+
+    runs-on: [self-hosted, vclust, '${{ matrix.machine }}']  
+
+    steps:
+    - name: make clean
+      run: make clean
+      continue-on-error: true
+    - name: make
+      run: make -j32 CXX=g++-12 STATIC_LINK=true LEIDEN=${{ matrix.leiden }}
+    - name: tar artifacts
+      run: tar -cvzf vclust.tar.gz vclust.py test.py LICENSE example bin
+
+
+########################################################################################
+  help:
+    name: Print usage
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [x64_linux, x64_mac, arm64_linux, arm64_mac]
+    runs-on: [self-hosted, vclust, '${{ matrix.machine }}']
+
+    steps:
+    - name: help
+      run: python3 vclust.py
+
+########################################################################################
+  upload:
+    name: Upload
+    needs: help
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [x64_linux, x64_mac, arm64_linux, arm64_mac]
+    runs-on: [self-hosted, vclust, '${{ matrix.machine }}']
+
+    steps:
+    - name: deploy
+      uses: actions/upload-release-asset@v1.0.1
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      with:
+        upload_url: ${{ github.event.release.upload_url }}
+        asset_path: ./vclust.tar.gz
+        asset_name: vclust-${{ github.event.release.tag_name }}-${{matrix.machine}}.tar.gz
+        asset_content_type: application/gzip

.github/workflows/main.yml

@@ -16,7 +16,7 @@ jobs:
       fail-fast: false
       matrix:
         machine: [ubuntu-latest, macOS-12]
-        compiler: [g++-11]
+        compiler: [g++-12]
     runs-on: ['${{ matrix.machine }}']   
 
     steps:

.gitmodules

@@ -1,7 +1,6 @@
 [submodule "3rd_party/kmer-db"]
 	path = 3rd_party/kmer-db
 	url = https://github.com/refresh-bio/kmer-db.git
-	branch = develop
 [submodule "3rd_party/lz-ani"]
 	path = 3rd_party/lz-ani
 	url = https://github.com/refresh-bio/LZ-ANI.git

test.py

@@ -13,6 +13,7 @@
 FASTA_DIR = DATA_DIR / 'fna'
 FASTA_FILE = DATA_DIR / 'multifasta.fna'
 ANI_FILE = DATA_DIR / 'output'/ 'ani.tsv'
+ALN_FILE = DATA_DIR / 'output' / 'ani.aln.tsv'
 IDS_FILE = DATA_DIR / 'output' / 'ani.ids.tsv'
 FLTR_FILE = DATA_DIR / 'output' / 'fltr.txt'
 
@@ -170,7 +171,7 @@ def test_align_default(test_dir, input, params):
     ref_pairs = {
         ('NC_010807.ref', 'NC_010807.alt1'): 0.99753,
         ('NC_010807.ref', 'NC_010807.alt2'): 0.98985,
-        ('NC_010807.ref', 'NC_010807.alt3'): 0.98414,
+        ('NC_010807.ref', 'NC_010807.alt3'): 0.98384,
         ('NC_005091.ref', 'NC_005091.alt1'): 0.97161,
         ('NC_005091.ref', 'NC_005091.alt2'): 0.96707,
         ('NC_025457.ref', 'NC_025457.alt1'): 0.80607,
@@ -188,7 +189,7 @@ def test_align_default(test_dir, input, params):
             pairs[(id1, id2)] = tani
     for ref_pair, ref_tani in ref_pairs.items():
         tani = pairs[ref_pair]
-        assert abs(tani - ref_tani) < 0.03
+        assert abs(tani - ref_tani) < 0.007
 
 
 @pytest.mark.parametrize('outfmt,ref_cols',[
@@ -214,6 +215,35 @@ def test_align_outfmt(test_dir, outfmt, ref_cols):
         assert cols == ref_cols
 
 
+@pytest.mark.parametrize('input,params',[
+    (FASTA_DIR, []),
+    (FASTA_FILE, []),
+])
+def test_align_alignments(test_dir, input, params):
+    out_file = test_dir.joinpath('ani.tsv')
+    out_aln_file = test_dir.joinpath('ani.aln.tsv')
+    cmd = [
+        f'{VCLUST.resolve()}',
+        'align',
+        '-i',
+        f'{input}',
+        '-o',
+        f'{out_file}',
+        '--out-aln',
+        f'{out_aln_file}',
+
+    ]
+    p = subprocess.run(cmd)
+    assert p.returncode == 0
+    assert p.stderr == None
+    assert out_aln_file.exists()
+    assert out_aln_file.stat().st_size
+    with open(out_aln_file) as fh:
+        header = fh.readline().split()
+        assert len(header) == 10
+        assert fh.readlines()
+
+
 @pytest.mark.parametrize('input,params',[
     (FASTA_DIR, []),
     (FASTA_FILE, []),
@@ -288,7 +318,12 @@ def test_cluster_algorithm(test_dir, algorithm):
     assert out_file.stat().st_size
 
 
-def test_cluster_algorithm_leiden(test_dir):
+@pytest.mark.parametrize('params',[
+    ([]),
+    (['--leiden-resolution', '0.8', '--leiden-iterations', '3']),
+    (['--leiden-resolution', '0.8', '--leiden-beta', '0.001']),
+])
+def test_cluster_algorithm_leiden(test_dir, params):
     out_file = test_dir / 'clusters.tsv'
     cmd = [
         f'{VCLUST.resolve()}',
@@ -306,6 +341,7 @@ def test_cluster_algorithm_leiden(test_dir):
         '--tani',
         '0.95',
     ]
+    cmd.extend(params)
     p = subprocess.run(cmd, 
         stdout=subprocess.DEVNULL, 
         stderr=subprocess.DEVNULL)