Checkpoint

.github/workflows/copilot-setup-steps.yml

@@ -36,5 +36,8 @@ jobs:
       - name: Create all conda envs
         shell: bash -l {0}
         run: |
-          conda env create -f envs/env_runner.yaml
+          conda env create -f envs/env.yaml
+          conda activate tcrtrifold-experiments
+          pip install -e .
+          conda deactivate
           conda env create -f envs/nf-core.yaml

.gitignore

@@ -123,5 +123,4 @@ tmp
 
 # data
 data/*
-!data/test
-!data/*.dvc
+logs

conf/af3.config

@@ -0,0 +1,56 @@
+
+
+params {
+    compress_inf = true
+    seeds = "1,2,3,4,5"
+    check_inf_exists = true
+    force_update_msa = false
+    save_embeddings = false
+}
+
+process {
+    withLabel: process_local {
+        executor = 'local'
+    }
+}
+
+profiles {
+    gemini {
+        params {
+            af3_model_dir = "/ref_genomes/alphafold/alphafold3/models"
+            af3_db_dir = "/ref_genomes/alphafold/alphafold3/"
+            msa_cache_dir = "/tgen_labs/altin/alphafold3/msa"
+        }
+
+        process {
+            withLabel: alphafold3_inference {
+                executor = 'slurm'
+                queue = 'gpu-a100'
+                cpus = '8'
+                memory = '64GB'
+                clusterOptions = '--nodes=1 --ntasks=1 --gres=gpu:1 --time=2:00:00'
+                container = '/tgen_labs/altin/alphafold3/containers/alphafold_3.0.1.sif'
+                containerOptions = '--nv -B /scratch,/tgen_labs,/ref_genomes --cleanenv'
+                beforeScript = 'module load singularity'
+            }
+
+            withLabel: alphafold3_msa {
+                executor = 'slurm'
+                queue = 'compute'
+                cpus = '8'
+                clusterOptions = '--nodes=1 --ntasks=1 --time=24:00:00'
+                container = '/tgen_labs/altin/alphafold3/containers/alphafold_3.0.1.sif'
+                containerOptions = '--nv -B /scratch,/tgen_labs,/ref_genomes --cleanenv'
+                beforeScript = 'module load singularity && export SINGULARITYENV_TMPDIR=/scratch/$USER/tmp'
+            }
+
+        }
+
+        singularity {
+            enabled = true
+        }
+
+    }
+
+
+}

conf/boltz.config

@@ -0,0 +1,29 @@
+
+profiles {
+
+    gemini {
+        params {
+            boltz_cache = "/tgen_labs/altin/boltz/.boltz"
+        }
+
+        process {
+
+            withLabel: boltz_local {
+                executor = 'local'
+                conda = "envs/boltz.yaml"
+            }
+
+            withLabel: boltz_gpu {
+                queue = { task.attempt == 1 ? 'gpu-v100' : 'gpu-a100' }
+                cpus ='8'
+                clusterOptions = '--nodes=1 --ntasks=1 --gres=gpu:1 --time=3:00:00'
+                memory = '64GB'
+                executor = "slurm"
+
+                errorStrategy = "retry"
+                maxRetries = 1
+                conda = "envs/boltz.yaml"
+            }
+        }
+    }
+}

workflows/nextflow.config → conf/local.config

@@ -1,25 +1,12 @@
-manifest.name = "Lawson Woods"
-manifest.version = "0.1.0"
-manifest.description = "Predicting TCR:pMHC binding status using AF3"
-
-nextflow.enable.moduleBinaries = true
-conda.enabled = true
-conda.createTimeout = '1 h'
-
-plugins {
-    id 'nf-parquet'
-}
-
-includeConfig "subworkflows/tgen/af3/nextflow.config"
-
-params.data_dir = "${launchDir}/data"
 
 profiles {
 
-    standard {
+    gemini {
+
+        workDir = "/scratch/$USER/work"
+
         params {
             imgt_hla_path = "/tgen_labs/altin/alphafold3/IMGTHLA"
-            msa_dir = "/tgen_labs/altin/alphafold3/msa"
             boltz_cache = "/tgen_labs/altin/boltz/.boltz"
         }
 
@@ -28,9 +15,18 @@ profiles {
                 executor = 'local'
             }
 
+            withLabel: mmseqs_heavy {
+                executor = 'slurm'
+                queue = 'compute'
+                cpus = '8'
+                memory = '64GB'
+                clusterOptions = '--time=1:00:00'
+                conda = "envs/mmseqs.yaml"
+            }
+
             withLabel: tcrtrifold_local {
                 executor = "local"
-                conda = "envs/env.yaml"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
             }
 
             withLabel: tcrtrifold_heavy {
@@ -40,7 +36,16 @@ profiles {
                 memory = '64GB'
                 executor = "slurm"
                 clusterOptions = '--time=8:00:00'
-                conda = "envs/env.yaml"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
+            }
+
+            withLabel: tcrtrifold_gpu {
+                queue = 'gpu-a100'
+                cpus ='8'
+                clusterOptions = '--nodes=1 --ntasks=1 --gres=gpu:1 --time=8:00:00'
+                memory = '64GB'
+                executor = "slurm"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
             }
 
             withLabel: tcrtrifold_very_heavy {
@@ -50,7 +55,7 @@ profiles {
                 memory = '64GB'
                 executor = "slurm"
                 clusterOptions = '--time=2-00:00:00'
-                conda = "envs/env.yaml"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
             }
 
             withLabel: tcrdock {
@@ -59,33 +64,25 @@ profiles {
                 cpus = '8'
                 memory = '64GB'
                 executor = "slurm"
-                clusterOptions = '--time=2-00:00:00'
+                clusterOptions = '--time=12:00:00'
                 conda = "envs/tcrdock.yaml"
             }
 
-            withLabel: boltz_local {
-                executor = 'local'
-                conda = "envs/boltz.yaml"
-            }
-
-            withLabel: boltz_gpu {
-                queue = { task.attempt == 1 ? 'gpu-v100' : 'gpu-a100' }
-                cpus ='8'
-                clusterOptions = '--nodes=1 --ntasks=1 --gres=gpu:1 --time=3:00:00'
-                memory = '64GB'
-                executor = "slurm"
+            // withLabel: extract_feat {
+            //     executor = 'slurm'
+            //     queue = "compute"
+            //     cpus = "8"
+            //     memory = "64 GB"
+            //     clusterOptions = "--time=8:00:00"
+            //     conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
+            // }
 
-                errorStrategy = "retry"
-                maxRetries = 1
-                conda = "envs/boltz.yaml"
-            }
         }
     }
 
     gh_runner {
         params {
             imgt_hla_path = null
-            msa_dir = "${launchDir}/data/test/msa"
         }
 
         process {
@@ -95,28 +92,16 @@ profiles {
 
             withLabel: tcrtrifold_local {
                 executor = "local"
-                conda = "envs/env_runner.yaml"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
             }
 
             withLabel: tcrtrifold_heavy {
                 executor = "local"
-                conda = "envs/env_runner.yaml"
+                conda = "$HOME/miniconda3/envs/tcrtrifold-experiments"
             }
         }
     }
 
 
 
 }
-
-
-
-process {
-    withLabel: extract_feat {
-        executor = 'slurm'
-        queue = "compute"
-        cpus = "8"
-        memory = "64 GB"
-        clusterOptions = "--time=8:00:00"
-    }
-}

envs/env.yaml

@@ -6,6 +6,7 @@ dependencies:
   - pip
   - python
   - ipykernel
+
   # to enable pip install of anarci
   - hmmer=3.3.2
   # for pandas interop
@@ -14,7 +15,6 @@ dependencies:
 
   - matplotlib
   - umap-learn
-  - pytorch
 
   - anarci
   - editdistance
@@ -30,6 +30,6 @@ dependencies:
 
   - pip:
     - git+https://github.com/ljwoods2/mdaf3.git@main
-    - -e /tgen_labs/altin/alphafold3/workspace/tcrtrifold-experiments
     - tcrdist3
-    - torch_geometric
+    - torch_geometric
+    - torch

envs/tcrdock.yaml

@@ -5,7 +5,8 @@ channels:
 dependencies:
   - pip
   - python
-  - blast
+  - blast<2.17
+  - anarci
 
 
   - pip:

main.nf

@@ -0,0 +1 @@
+// dummy file- needed for nextlow