PIPE-89-genophase-script-donor (#170)

* added donor_id generation * print instead of warning * if else fixed * analysis needs to be hic * raise error instead of print * run tox in python3.9 * no shell-lint in pre-commit * run python 3.9.10 in circleci * pysam 0.19.1 Co-authored-by: Otto Jolanki <[email protected]>
ENCODE-DCC · Oct 3, 2022 · 511636c · 511636c
1 parent 1282555
commit 511636c
Show file tree

Hide file tree

Showing 4 changed files with 41 additions and 24 deletions.
diff --git a/.circleci/config.yml b/.circleci/config.yml
@@ -32,17 +32,17 @@ machine_defaults: &machine_defaults
     image: ubuntu-2004:202201-02
   working_directory: ~/ENCODE-DCC/hic-pipeline
 
-py37: &py37
+py39: &py39
   docker:
-    - image: circleci/python:3.7.4-stretch
+    - image: circleci/python:3.9
   working_directory: ~/ENCODE-DCC/hic-pipeline
 
 commands:
   run_tox:
     description: Install and run tox with a given environment
     parameters:
       toxenv:
-        description: The name of the environment as per tox.ini, e.g. py37 or lint
+        description: The name of the environment as per tox.ini, e.g. py39 or lint
         type: string
       extra_args:
         description: Extra arguments that are consumed only when running pytest
@@ -101,19 +101,19 @@ commands:
 
 jobs:
   lint:
-    <<: *py37
+    <<: *py39
     steps:
       - checkout
       - install_shellcheck
       - run_tox:
           toxenv: lint
 
-  test_py37:
-    <<: *py37
+  test_py39:
+    <<: *py39
     steps:
       - checkout
       - run_tox:
-          toxenv: py37
+          toxenv: py39
 
   test_wdl:
     <<: *machine_defaults
@@ -142,8 +142,8 @@ jobs:
       - run:
           name: Run pytest-workflow WDL tests using tox
           command: |
-            pyenv install 3.7.0
-            pyenv global 3.7.0
+            pyenv install 3.9.10
+            pyenv global 3.9.10
             pip install --upgrade pip
             pip install tox
             tox -e wdl -- --kwd --tag << parameters.tag >> --wt << parameters.num_parallel_workflows >>
@@ -223,7 +223,7 @@ workflows:
           dockerfile_directory: docker/delta
           cache_tag: << pipeline.parameters.delta_cache_tag >>
       - lint
-      - test_py37:
+      - test_py39:
           requires:
             - lint
       - test_wdl:

diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -5,7 +5,7 @@ repos:
     rev: 22.3.0
     hooks:
       - id: black
-        language_version: python3.7
+        language_version: python3.9
 
   - repo: https://github.com/asottile/seed-isort-config
     rev: v1.9.2
@@ -16,12 +16,7 @@ repos:
     rev: v4.3.21
     hooks:
       - id: isort
-        language_version: python3.7
-
-  - repo: https://github.com/detailyang/pre-commit-shell
-    sha: v1.0.6
-    hooks:
-      - id: shell-lint
+        language_version: python3.9
 
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v2.2.3

diff --git a/scripts/make_genophase_input_json_from_portal.py b/scripts/make_genophase_input_json_from_portal.py
@@ -25,7 +25,8 @@ def main():
         auth=auth,
         use_submitted_file_names=args.use_submitted_file_names,
     )
-    input_json = _get_input_json(bams=bams)
+    donor_id = _get_donor_id_from_experiment(accessions=args.accessions, auth=auth)
+    input_json = _get_input_json(bams=bams, donor_id=donor_id)
     _write_json_to_file(input_json, args.outfile)
 
 
@@ -48,6 +49,7 @@ def _get_bams_from_experiment(accessions, auth=None, use_submitted_file_names=Fa
             if (
                 a["status"] in _ALLOWED_STATUSES
                 and a["lab"] == "/labs/encode-processing-pipeline/"
+                and a["pipelines"][0] == "/pipelines/ENCPL839OAB/"
             ):
                 analysis = a
                 break
@@ -66,11 +68,31 @@ def _get_bams_from_experiment(accessions, auth=None, use_submitted_file_names=Fa
     return bams
 
 
-def _get_input_json(bams):
+def _get_donor_id_from_experiment(accessions, auth=None):
+    donor_id = None
+    for accession in accessions:
+        experiment = _get_experiment(accession, auth=auth)
+        if donor_id is None:
+            donor_id = experiment["replicates"][0]["library"]["biosample"]["donor"][
+                "accession"
+            ]
+        else:
+            if (
+                donor_id
+                != experiment["replicates"][0]["library"]["biosample"]["donor"][
+                    "accession"
+                ]
+            ):
+                raise ValueError("These datasets do not belong to the same donor.")
+    return donor_id
+
+
+def _get_input_json(bams, donor_id):
     input_json = {
         "genophase.bams": bams,
         "genophase.gatk_bundle_tar": _REFERENCE_FILES["gatk_bundle_tar"],
         "genophase.reference_fasta": _REFERENCE_FILES["reference_fasta"],
+        "genophase.donor_id": donor_id,
     }
     return input_json
 

diff --git a/tox.ini b/tox.ini
@@ -1,5 +1,5 @@
 [tox]
-envlist = lint,py37,wdl,coverage-report
+envlist = lint,py39,wdl,coverage-report
 isolated_build = True
 skipsdist = True
 
@@ -10,22 +10,22 @@ deps =
     -rrequirements-scripts.txt
 
 [testenv:wdl]
-basepython = python3.7
+basepython = python3.9
 commands = python -m pytest --ignore=tests/python --symlink {posargs}
 deps =
     caper==1.6.3
-    pysam==0.15.4
+    pysam==0.19.1
     pytest-workflow==1.5.0
     pytest==6.2.2
 
 [testenv:lint]
-basepython = python3.7
+basepython = python3.9
 skip_install = true
 deps = pre-commit
 commands = pre-commit run --all-files
 
 [testenv:coverage-report]
-basepython = python3.7
+basepython = python3.9
 commands = pytest --cov-report term-missing --cov=hic-pipeline
 
 [flake8]