Rename task-dge-perturbation-prediction to `task_perturbation_predi…

…ction` (openproblems-bio#66)

* update task info

* update project config

* rename task

* move files

* add namespace to api

* update readme

* update spec

* simplify component

CHANGELOG.md

@@ -1,10 +1,10 @@
-# task-dge-perturbation-prediction 0.1.0
+# task_perturbation_prediction 1.0.0
 
-Initial release of the DGE Perturbation Prediction task. Initial components:
+Initial release of the Perturbation Prediction task. Initial components:
 
-* `src/task/process_dataset`: Compute the DGE data from the raw single-cell counts using Limma.
-* `src/task/control_methods`: Baseline control methods: sample, ground_truth, zeros, mean_across_celltypes, mean_across_compounds, mean_outcome.
-* `src/task/methods`: DGE perturbation prediction methods: random_forest.
-* `src/task/metrics`: Evaluation metrics: mean_rowwise_error.
+* `src/process_dataset`: Compute the DGE data from the raw single-cell counts using Limma.
+* `src/control_methods`: Baseline control methods: sample, ground_truth, zeros, mean_across_celltypes, mean_across_compounds, mean_outcome.
+* `src/methods`: Perturbation prediction methods: jn_ap_op2, lgc_ensemble, nn_retraining_with_pseudolabels, pyboost, scape, transformer_ensemble.
+* `src/metrics`: Evaluation metrics: mean_rowwise_error, mean_rowwise_correlation.
 
 

_viash.yaml

@@ -7,8 +7,8 @@ config_mods: |
   .functionality.version := 'dev'
   .functionality.arguments[.multiple == true].multiple_sep := ';'
   .platforms[.type == 'docker'].target_registry := 'ghcr.io'
-  .platforms[.type == 'docker'].target_organization := 'openproblems-bio/task-dge-perturbation-prediction'
-  .platforms[.type == 'docker'].target_image_source := 'https://github.com/openproblems-bio/task-dge-perturbation-prediction'
+  .platforms[.type == 'docker'].target_organization := 'openproblems-bio/task_perturbation_prediction'
+  .platforms[.type == 'docker'].target_image_source := 'https://github.com/openproblems-bio/task_perturbation_prediction'
   .platforms[.type == "nextflow"].directives.tag := "$id"
   .platforms[.type == "nextflow"].auto.simplifyOutput := false
   .platforms[.type == "nextflow"].config.labels := { lowmem : "memory = 20.Gb", midmem : "memory = 50.Gb", highmem : "memory = 100.Gb", lowcpu : "cpus = 5", midcpu : "cpus = 15", highcpu : "cpus = 30", lowtime : "time = 1.h", midtime : "time = 4.h", hightime : "time = 8.h", veryhightime : "time = 24.h" }

scripts/add_a_method.sh

@@ -15,25 +15,25 @@ viash run src/common/create_component/config.vsh.yaml -- \
   --language "$method_lang" \
   --name "$method_id"
 
-# TODO: fill in required fields in src/task/methods/foo/config.vsh.yaml
-# TODO: edit src/task/methods/foo/script.py/R
+# TODO: fill in required fields in src/methods/foo/config.vsh.yaml
+# TODO: edit src/methods/foo/script.py/R
 
 # test the component
-viash test src/task/methods/$method_id/config.vsh.yaml
+viash test src/methods/$method_id/config.vsh.yaml
 
 # rebuild the container (only if you change something to the docker platform)
 # You can reduce the memory and cpu allotted to jobs in _viash.yaml by modifying .platforms[.type == "nextflow"].config.labels
-viash run src/task/methods/$method_id/config.vsh.yaml -- \
+viash run src/methods/$method_id/config.vsh.yaml -- \
   ---setup cachedbuild ---verbose
 
 # run the method (using h5ad as input)
-viash run src/task/methods/$method_id/config.vsh.yaml -- \
+viash run src/methods/$method_id/config.vsh.yaml -- \
   --de_train_h5ad "resources/neurips-2023-kaggle/2023-09-12_de_by_cell_type_train.h5ad" \
   --id_map "resources/neurips-2023-kaggle/id_map.csv" \
   --output "output/prediction.h5ad"
 
 # run evaluation metric
-viash run src/task/metrics/mean_rowwise_error/config.vsh.yaml -- \
+viash run src/metrics/mean_rowwise_error/config.vsh.yaml -- \
   --de_test_h5ad "resources/neurips-2023-kaggle/de_test.h5ad" \
   --prediction "output/prediction.h5ad" \
   --output "output/score.h5ad"

scripts/generate_kaggle_resources.sh

@@ -18,7 +18,7 @@ if [[ ! -f "$OUT/2023-09-12_de_by_cell_type_test.h5ad" ]]; then
     "import anndata as ad; ad.read_h5ad('$OUT/2023-09-12_de_by_cell_type_train.h5ad').write_h5ad('$OUT/2023-09-12_de_by_cell_type_train.h5ad', compression='gzip')"
 fi
 
-viash run src/task/process_dataset/convert_kaggle_h5ad_to_parquet/config.vsh.yaml -- \
+viash run src/process_dataset/convert_kaggle_h5ad_to_parquet/config.vsh.yaml -- \
   --input_train "$OUT/2023-09-12_de_by_cell_type_train.h5ad" \
   --input_test "$OUT/2023-09-12_de_by_cell_type_test.h5ad" \
   --input_single_cell_h5ad "resources/neurips-2023-raw/sc_counts.h5ad" \
@@ -34,14 +34,14 @@ viash run src/task/process_dataset/convert_kaggle_h5ad_to_parquet/config.vsh.yam
   --dataset_organism homo_sapiens
 
 echo ">> Run method"
-viash run src/task/control_methods/mean_across_compounds/config.vsh.yaml -- \
+viash run src/control_methods/mean_across_compounds/config.vsh.yaml -- \
   --de_train_h5ad "$OUT/de_train.h5ad" \
   --de_test_h5ad "$OUT/de_test.h5ad" \
   --id_map "$OUT/id_map.csv" \
   --output "$OUT/prediction.h5ad"
 
 echo ">> Run metric"
-viash run src/task/metrics/mean_rowwise_error/config.vsh.yaml -- \
+viash run src/metrics/mean_rowwise_error/config.vsh.yaml -- \
   --prediction "$OUT/prediction.h5ad" \
   --de_test_h5ad "$OUT/de_test.h5ad" \
   --output "$OUT/score.h5ad"

scripts/generate_resources.sh

@@ -32,14 +32,14 @@ nextflow run \
   --publish_dir "$OUT"
 
 echo ">> Run method"
-viash run src/task/control_methods/mean_across_compounds/config.vsh.yaml -- \
+viash run src/control_methods/mean_across_compounds/config.vsh.yaml -- \
   --de_train_h5ad "$OUT/de_train.h5ad" \
   --de_test_h5ad "$OUT/de_test.h5ad" \
   --id_map "$OUT/id_map.csv" \
   --output "$OUT/prediction.h5ad"
 
 echo ">> Run metric"
-viash run src/task/metrics/mean_rowwise_error/config.vsh.yaml -- \
+viash run src/metrics/mean_rowwise_error/config.vsh.yaml -- \
   --prediction "$OUT/prediction.h5ad" \
   --de_test_h5ad "$OUT/de_test.h5ad" \
   --output "$OUT/score.h5ad"

scripts/render_readme.sh

@@ -5,7 +5,7 @@ set -e
 [[ ! -d ../openproblems-v2 ]] && echo "You need to clone the openproblems-v2 repository next to this repository" && exit 1
 
 ../openproblems-v2/target/docker/common/create_task_readme/create_task_readme \
-  --task "dge_perturbation_prediction" \
-  --task_dir "src/task" \
-  --github_url "https://github.com/openproblems-bio/task-dge-perturbation-prediction/tree/main/" \
+  --task "perturbation_prediction" \
+  --task_dir "src" \
+  --github_url "https://github.com/openproblems-bio/task_perturbation_prediction/tree/main/" \
   --output "README.md"

scripts/run_benchmark_tw.sh

@@ -2,7 +2,7 @@
 
 RUN_ID="run_$(date +%Y-%m-%d_%H-%M-%S)"
 resources_dir="s3://openproblems-bio/public/neurips-2023-competition/workflow-resources"
-publish_dir="s3://openproblems-data/resources/dge_perturbation_prediction/results/${RUN_ID}"
+publish_dir="s3://openproblems-data/resources/perturbation_prediction/results/${RUN_ID}"
 
 cat > /tmp/params.yaml << HERE
 param_list:
@@ -20,7 +20,7 @@ output_state: "state.yaml"
 publish_dir: "$publish_dir"
 HERE
 
-tw launch https://github.com/openproblems-bio/task-dge-perturbation-prediction.git \
+tw launch https://github.com/openproblems-bio/task_perturbation_prediction.git \
   --revision main_build \
   --pull-latest \
   --main-script target/nextflow/workflows/run_benchmark/main.nf \

scripts/run_benchmark_tw_traens.sh

@@ -4,7 +4,7 @@
 
 RUN_ID="traens_$(date +%Y-%m-%d_%H-%M-%S)"
 resources_dir="s3://openproblems-bio/public/neurips-2023-competition/workflow-resources"
-publish_dir="s3://openproblems-data/resources/dge_perturbation_prediction/results/${RUN_ID}"
+publish_dir="s3://openproblems-data/resources/perturbation_prediction/results/${RUN_ID}"
 
 cat > /tmp/params.yaml << HERE
 param_list:
@@ -18,8 +18,8 @@ output_state: "state.yaml"
 publish_dir: "$publish_dir"
 HERE
 
-tw launch https://github.com/openproblems-bio/task-dge-perturbation-prediction.git \
-  --revision fix_trafo_ens_build \
+tw launch https://github.com/openproblems-bio/task_perturbation_prediction.git \
+  --revision suggestions_elior_build \
   --pull-latest \
   --main-script target/nextflow/workflows/run_benchmark/main.nf \
   --workspace 53907369739130 \

scripts/run_layert_tw.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 RUN_ID="layert_$(date +%Y-%m-%d_%H-%M-%S)"
-publish_dir="s3://openproblems-data/resources/dge_perturbation_prediction/results/${RUN_ID}"
+publish_dir="s3://openproblems-data/resources/perturbation_prediction/results/${RUN_ID}"
 
 cat > /tmp/params.yaml << HERE
 id: dge_perturbation_task
@@ -12,7 +12,7 @@ rename_keys: "de_train_h5ad:de_train_h5ad,de_test_h5ad:de_test_h5ad,id_map:id_ma
 settings: '{"layer": "t"}'
 HERE
 
-tw launch https://github.com/openproblems-bio/task-dge-perturbation-prediction.git \
+tw launch https://github.com/openproblems-bio/task_perturbation_prediction.git \
   --revision main_build \
   --pull-latest \
   --main-script target/nextflow/workflows/run_benchmark/main.nf \

scripts/run_stability_tw.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 RUN_ID="stability_$(date +%Y-%m-%d_%H-%M-%S)"
-publish_dir="s3://openproblems-data/resources/dge_perturbation_prediction/results/${RUN_ID}"
+publish_dir="s3://openproblems-data/resources/perturbation_prediction/results/${RUN_ID}"
 
 cat > /tmp/params.yaml << HERE
 id: neurips-2023-data
@@ -11,7 +11,7 @@ output_state: "state.yaml"
 publish_dir: "$publish_dir"
 HERE
 
-tw launch https://github.com/openproblems-bio/task-dge-perturbation-prediction.git \
+tw launch https://github.com/openproblems-bio/task_perturbation_prediction.git \
   --revision main_build \
   --pull-latest \
   --main-script target/nextflow/workflows/run_stability_analysis/main.nf \

scripts/sync_results.sh

@@ -1,18 +1,18 @@
 #!/bin/bash
 
 aws s3 sync \
-  s3://openproblems-data/resources/dge_perturbation_prediction/results/ \
+  s3://openproblems-data/resources/perturbation_prediction/results/ \
   output/benchmark_results/ \
   --delete --dryrun
 
 # sync back modified results
 aws s3 sync \
   output/benchmark_results/ \
-  s3://openproblems-data/resources/dge_perturbation_prediction/results/ \
+  s3://openproblems-data/resources/perturbation_prediction/results/ \
   --delete --dryrun
 
 # sync one run
 runid=run_2024-06-01_00-03-09; aws s3 sync \
   output/benchmark_results/${runid}/ \
-  s3://openproblems-data/resources/dge_perturbation_prediction/results/${runid}/ \
-  --delete --dryrun
+  s3://openproblems-data/resources/perturbation_prediction/results/${runid}/ \
+  --delete --dryrun

src/task/api/comp_method_notest.yaml → src/api/comp_method_notest.yaml

@@ -26,8 +26,7 @@ functionality:
       required: true
       direction: output
     - name: "--output_model"
-      type: "file"
-      description: "Optional model output. If no value is passed, the model will be removed at the end of the run."
+      __merge__: file_model.yaml
       direction: output
       required: false
       must_exist: false

src/task/api/comp_process_dataset.yaml → src/api/comp_process_dataset.yaml

@@ -1,4 +1,5 @@
 functionality:
+  namespace: process_dataset
   info:
     type: process_dataset
     type_info:
@@ -25,5 +26,4 @@ functionality:
       __merge__: file_id_map.yaml
       required: true
       direction: output
-      default: id_map.csv
-  test_resources: []
+      default: id_map.csv

src/api/file_model.yaml

@@ -0,0 +1,6 @@
+type: file
+example: resources/neurips-2023-data/model/
+info:
+  label: Model
+  summary: "Optional model output. If no value is passed, the model will be removed at the end of the run."
+  file_type: directory

src/task/api/task_info.yaml → src/api/task_info.yaml

@@ -1,5 +1,5 @@
-name: dge_perturbation_prediction
-label: DGE Perturbation Prediction
+name: perturbation_prediction
+label: Perturbation Prediction
 summary: Predicting how small molecules change gene expression in different cell types.
 readme: |
   ## Installation
@@ -17,9 +17,9 @@ readme: |
   To get started, you can run the following commands:
 
   ```bash
-  git clone git@github.com:openproblems-bio/task-dge-perturbation-prediction.git
+  git clone git@github.com:openproblems-bio/task_perturbation_prediction.git
 
-  cd task-dge-perturbation-prediction
+  cd task_perturbation_prediction
   
   # download resources
   scripts/download_resources.sh
@@ -99,3 +99,43 @@ authors:
     info:
       github: rcannood
       orcid: "0000-0003-3641-729X"
+  - name: Daniel Burkhardt
+    roles: [ author ]
+    info:
+      github: dburkhardt
+      orcid: 0000-0001-7744-1363
+  - name: Malte D. Luecken
+    roles: [ author ]
+    info:
+      github: LuckyMD
+      orcid: 0000-0001-7464-7921
+  - name: Tin M. Tunjic
+    roles: [ contributor ]
+    info:
+      github: ttunja
+      orcid: 0000-0001-8842-6548
+  - name: Mengbo Wang
+    roles: [ contributor ]
+    info:
+      github: wangmengbo
+      orcid: 0000-0002-0266-9993
+  - name: Andrew Benz
+    roles: [ author ]
+    info:
+      github: andrew-benz
+      orcid: 0009-0002-8118-1861
+  - name: Tianyu Liu
+    roles: [ contributor ]
+    info:
+      github: HelloWorldLTY
+      orcid: 0000-0002-9412-6573
+  - name: Jalil Nourisa
+    roles: [ contributor ]
+    info:
+      github: janursa
+      orcid: 0000-0002-7539-4396
+  - name: Rico Meinl
+    roles: [ contributor ]
+    info:
+      github: ricomnl
+      orcid: 0000-0003-4356-6058

src/common/create_component/config.vsh.yaml

@@ -24,7 +24,7 @@ functionality:
       direction: output
       # required: true
       description: Path to the component directory. Suggested location is `src/<TASK>/<TYPE>s/<NAME>`.
-      default: src/task/methods/${VIASH_PAR_NAME}
+      default: src/methods/${VIASH_PAR_NAME}
     - type: file
       name: --api_file
       description: |
@@ -33,7 +33,7 @@ functionality:
         to manually specify a different API file to inherit from.
       must_exist: false
       # required: true
-      default: src/task/api/comp_method.yaml
+      default: src/api/comp_method.yaml
     - type: file
       name: --viash_yaml
       description: |

src/common/create_component/script.py

@@ -6,12 +6,12 @@
 
 ## VIASH START
 par = {
-  "task": "DGE Perturbation Prediction",
+  "task": "Perturbation Prediction",
   "type": "method",
   "language": "python",
   "name": "new_comp",
-  "output": "src/task/method/new_comp",
-  "api_file": "src/task/api/comp_method.yaml",
+  "output": "src/method/new_comp",
+  "api_file": "src/api/comp_method.yaml",
   "viash_yaml": "_viash.yaml"
 }
 ## VIASH END

src/task/methods/jn_ap_op2/config.vsh.yaml → src/methods/jn_ap_op2/config.vsh.yaml

@@ -30,7 +30,6 @@ functionality:
     - type: python_script
       path: script.py
     - path: helper.py
-    - path: ../../utils/anndata_to_dataframe.py
 platforms:
   - type: docker
     image: ghcr.io/openproblems-bio/base_pytorch_nvidia:1.0.4

src/task/methods/jn_ap_op2/script.py → src/methods/jn_ap_op2/script.py

@@ -20,18 +20,16 @@
     "submission_names": ["dl40"]
 }
 meta = {
-    "resources_dir": "src/task/methods/jn_ap_op2",
+    "resources_dir": "src/methods/jn_ap_op2",
 }
 ## VIASH END
 
 sys.path.append(meta["resources_dir"])
 
-from anndata_to_dataframe import anndata_to_dataframe
 from helper import plant_seed, MultiOutputTargetEncoder, train
 
 print('Reading input files', flush=True)
 de_train_h5ad = ad.read_h5ad(par["de_train_h5ad"])
-de_train = anndata_to_dataframe(de_train_h5ad, par["layer"])
 id_map = pd.read_csv(par["id_map"])
 
 gene_names = list(de_train_h5ad.var_names)
@@ -58,10 +56,10 @@
 
 print('Data location', flush=True)
 # Data location
-cell_types = de_train['cell_type']
-sm_names = de_train['sm_name']
+cell_types = de_train_h5ad.obs['cell_type'].astype(str)
+sm_names = de_train_h5ad.obs['sm_name'].astype(str)
 
-data = de_train.drop(columns=["cell_type", "sm_name", "sm_lincs_id", "SMILES", "split", "control"]).to_numpy(dtype=float)
+data = de_train_h5ad.layers[par["layer"]]
 
 print('Train model', flush=True)
 # ... train model ...

src/task/methods/lgc_ensemble_direct/script.py → src/methods/lgc_ensemble_direct/script.py

@@ -19,7 +19,7 @@
     "output_model": None
 }
 meta = {
-    "resources_dir": "src/task/methods/lgc_ensemble",
+    "resources_dir": "src/methods/lgc_ensemble",
     "temp_dir": "/tmp"
 }
 ## VIASH END