Initial commit

.github/workflows/main.yml

@@ -0,0 +1,73 @@
+name: GitHub Actions CI
+
+on:
+  push:
+    branches: [ main, master, develop ]
+    paths-ignore:
+      - '**.md'
+  workflow_dispatch:
+
+jobs:
+
+########################################################################################
+  make:
+    name: Make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [ubuntu-latest, macOS-12]
+        compiler: [g++-11]
+    runs-on: ['${{ matrix.machine }}']   
+
+    steps:
+    - uses: actions/checkout@v2
+      with:
+        submodules: recursive
+
+    - name: get tags
+      run: | 
+        cd ./libs/igraph
+        git fetch --tags origin
+    
+    - name: make
+      run: | 
+        make -j32 CXX=${{matrix.compiler}} 
+        cp ./clusty ./clusty-${{matrix.compiler}}
+        make clean
+    - name: tar artifacts
+      run: tar -cvf clusty.tar ./clusty-${{matrix.compiler}} ./test/cmp.py ./test/vir61.*
+
+    - uses: actions/upload-artifact@v2
+      with:
+        name: executable-artifact-${{ matrix.machine }}
+        path: ./clusty.tar    
+
+########################################################################################
+  vir61-linkage:
+    name: Vir61
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [ubuntu-latest, macOS-12]
+        compiler: [g++-11]
+        algo: [single, complete]
+        threshold: [95, 70]
+
+    runs-on: ['${{ matrix.machine }}']    
+
+    steps:
+    - uses: actions/download-artifact@v2
+      with:
+        name: executable-artifact-${{ matrix.machine }}
+        path: ./
+
+    - name: untar artifacts
+      run: |
+        tar -xf clusty.tar
+    
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives, reordered columns)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/vir61.list --similarity --min ani 0.${{matrix.threshold}} --id-cols id2 id1 --distance-col ani ./test/vir61.ani vir61.${{matrix.algo}}.${{matrix.threshold}} --out-representatives
+        python ./test/cmp.py vir61.${{matrix.algo}}.${{matrix.threshold}} ./test/vir61.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+  

.github/workflows/self-hosted.yml

@@ -0,0 +1,202 @@
+name: Self-hosted CI
+
+on:
+  pull_request:
+    branches: [main, master, develop]
+    paths-ignore:
+      - '**.md'
+  workflow_dispatch:
+
+
+jobs:
+
+  ########################################################################################
+  checkout:
+    name: Checkout
+    strategy:
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']
+
+    steps:
+    - uses: actions/checkout@v2
+      with:
+        submodules: recursive
+    - name: Get tags
+      run: | 
+        cd ./libs/igraph
+        git fetch --tags origin
+    
+
+  ########################################################################################
+  make:
+    name: Make
+    needs: checkout
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+        compiler: [g++-11]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']
+
+    steps:
+    - name: make
+      run: | 
+        make -j32 CXX=${{matrix.compiler}} 
+        cp ./clusty ./clusty-${{matrix.compiler}}
+        make clean
+
+  ########################################################################################
+  make-leiden:
+    name: make (Leiden)
+
+    needs: checkout
+    strategy:
+      fail-fast: false
+      matrix:
+        compiler: [g++-11]
+
+    runs-on: [self-hosted, clusty]
+
+    steps:
+    - name: make
+      run: | 
+        make -j32 CXX=${{matrix.compiler}} LEIDEN=true
+        cp ./clusty ./clusty-leiden-${{matrix.compiler}}
+        make clean
+  
+  ########################################################################################                 
+  toy:
+    name: Toy example
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+        compiler: [g++-11]
+        algo: [single]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']    
+
+    steps:
+    - name: ${{matrix.algo}} (no representatives, numeric ids, singletons in object file)
+      run: |
+        ./clusty-${{matrix.compiler}} --objects-file ./test/toy.ids.tsv --algo ${{matrix.algo}} --id-cols idx1 idx2 --distance-col tani --similarity --numeric-ids --min tani 0.95 ./test/toy.ani.tsv toy.${{matrix.algo}}.tsv
+        python3 ./test/cmp.py toy.${{matrix.algo}}.tsv ./test/toy.${{matrix.algo}}.tsv
+        
+  ########################################################################################        
+
+  vir61-linkage:
+    name: Vir61
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+        compiler: [g++-11]
+        algo: [single, complete]
+        threshold: [95, 70]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']    
+
+    steps:
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives, reordered columns)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/vir61.list --similarity --min ani 0.${{matrix.threshold}} --id-cols id2 id1 --distance-col ani ./test/vir61.ani vir61.${{matrix.algo}}.${{matrix.threshold}} --out-representatives
+        python3 ./test/cmp.py vir61.${{matrix.algo}}.${{matrix.threshold}} ./test/vir61.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+  
+  ########################################################################################   
+  ictv-linkage:
+    name: ICTV (single and complete versus python clusterings)
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+        compiler: [g++-11]
+        algo: [single, complete]
+        threshold: [95, 70]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']    
+
+    steps:
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (no representatives)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}
+        python3 ./test/cmp.py ictv.${{matrix.algo}}.${{matrix.threshold}} ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}.reps --out-representatives
+        python3 ./test/cmp.py ictv.${{matrix.algo}}.${{matrix.threshold}}.reps ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (no representatives, numeric ids)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.num ictv.${{matrix.algo}}.${{matrix.threshold}}.num --numeric-ids
+        python3 ./test/cmp.py ictv.${{matrix.algo}}.${{matrix.threshold}}.num ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives, numeric ids)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.num ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num --out-representatives --numeric-ids
+        python3 ./test/cmp.py ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.python.csv
+  
+  ########################################################################################   
+  ictv-all:
+    name: ICTV (all algos versus our clusterings)
+    needs: make
+    strategy:
+      fail-fast: false
+      matrix:
+        machine: [xeon, mac-i7, mac-m1]
+        compiler: [g++-11]
+        algo: [single, complete, uclust, set-cover, cd-hit]
+        threshold: [70]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']    
+
+    steps:
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (no representatives)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}} ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}.reps --out-representatives
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}}.reps ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives, numeric ids)
+      run: |
+        ./clusty-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.num ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num --out-representatives --numeric-ids
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.csv
+        
+  ########################################################################################   
+  ictv-leiden:
+    name: ICTV (leiden)
+    needs: make-leiden
+    strategy:
+      fail-fast: false
+      matrix:
+        compiler: [g++-11]
+        algo: [leiden]
+        threshold: [70]
+
+    runs-on: [self-hosted, clusty, '${{ matrix.machine }}']    
+
+    steps:
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (no representatives)
+      run: |
+        ./clusty-leiden-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}} ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives)
+      run: |
+        ./clusty-leiden-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.ani ictv.${{matrix.algo}}.${{matrix.threshold}}.reps --out-representatives
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}}.reps ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.csv
+        
+    - name: ${{matrix.algo}}, ${{matrix.threshold}} (with representatives, numeric ids)
+      run: |
+        ./clusty-keiden-${{matrix.compiler}} --algo ${{matrix.algo}} --objects-file ./test/ictv.list --similarity --min ani 0.${{matrix.threshold}} ./test/ictv.num ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num --out-representatives --numeric-ids
+        cmp ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.num ./test/ictv.${{matrix.algo}}.${{matrix.threshold}}.reps.csv
+  

.gitignore

@@ -0,0 +1,11 @@
+/src/x64
+/src/_release/
+/src/_debug/
+/src/.vs/rapid-cluster/v17
+/src/.vs/rapid-cluster/FileContentIndex
+/src/.vs/clusty/v17
+/src/.vs/clusty/FileContentIndex
+/libs/winflexbison
+/src/clusty/x64/Debug/clusty.tlog
+/src/clusty/x64/Debug
+/src

.gitmodules

@@ -0,0 +1,6 @@
+[submodule "libs/mimalloc"]
+	path = libs/mimalloc
+	url = https://github.com/refresh-bio-dependencies/mimalloc.git
+[submodule "libs/igraph"]
+	path = libs/igraph
+	url = https://github.com/refresh-bio-dependencies/igraph