PIP-883-mirna-template-tech-reps (#38)

* update template and add raw fastqs key for mirna

accession/__main__.py

@@ -6,7 +6,6 @@
 
 from accession import __version__
 from accession.accession import accession_factory
-from accession.analysis import Analysis, MetaData
 from accession.helpers import filter_outputs_by_path
 
 
@@ -75,12 +74,10 @@ def main(args=None):
     if args.filter_from_path:
         filter_outputs_by_path(args.filter_from_path)
         return
-    metadata = MetaData(args.accession_metadata)
-    analysis = Analysis(metadata)
     connection = Connection(args.server)
-    if all([args.pipeline_type, analysis, lab, award, connection]):
+    if all([args.pipeline_type, args.accession_metadata, lab, award, connection]):
         accessioner = accession_factory(
-            args.pipeline_type, analysis, connection, lab, award
+            args.pipeline_type, args.accession_metadata, connection, lab, award
         )
         accessioner.accession_steps()
         return

accession/accession.py

@@ -6,6 +6,7 @@
 from pathlib import Path
 from typing import Optional, Type
 
+from accession.analysis import Analysis, MetaData
 from accession.helpers import string_to_number
 from accession.quality_metric import QualityMetric
 
@@ -18,19 +19,26 @@ def __init__(self, path_to_accession_step_json):
         self._path_to_accession_step_json = path_to_accession_step_json
         self._steps = None
 
+    def _load_steps(self):
+        if self._steps:
+            return
+        with open(self.path_to_json) as fp:
+            self._steps = json.load(fp)
+
     @property
     def path_to_json(self):
         return self._path_to_accession_step_json
 
     @property
     def content(self):
-        if self._steps:
-            return self._steps["accession.steps"]
-        else:
-            with open(self.path_to_json) as fp:
-                self._steps = json.load(fp)
+        self._load_steps()
         return self._steps["accession.steps"]
 
+    @property
+    def raw_fastqs_keys(self) -> Optional[str]:
+        self._load_steps()
+        return self._steps.get("raw_fastqs_keys")
+
 
 class Accession(ABC):
     """docstring for Accession
@@ -657,7 +665,9 @@ def make_star_qc_metric(self, encode_bam_file, gs_file):
         return self.queue_qc(star_qc_metric, encode_bam_file, "star-quality-metric")
 
 
-def accession_factory(pipeline_type: str, *args, **kwargs) -> Accession:
+def accession_factory(
+    pipeline_type: str, accession_metadata: str, *args, **kwargs
+) -> Accession:
     """
     Matches against the user-specified pipeline_type string and returns an instance of
     the appropriate accession subclass. Usage of this factory has the nice effect of
@@ -680,4 +690,6 @@ def accession_factory(pipeline_type: str, *args, **kwargs) -> Accession:
         current_dir.parents[1] / "accession_steps" / f"{pipeline_type}_steps.json"
     )
     accession_steps = AccessionSteps(steps_json_path)
-    return selected_accession(accession_steps, *args, **kwargs)
+    metadata = MetaData(accession_metadata)
+    analysis = Analysis(metadata, raw_fastqs_keys=accession_steps.raw_fastqs_keys)
+    return selected_accession(accession_steps, analysis, *args, **kwargs)

accession/analysis.py

@@ -22,10 +22,13 @@ class Analysis(object):
         Args: metadata: MetaData object
     """
 
-    def __init__(self, metadata, auto_populate=True, backend=None):
+    def __init__(
+        self, metadata, raw_fastqs_keys=None, auto_populate=True, backend=None
+    ):
         self.files = []
         self.tasks = []
         self.metadata = metadata.content
+        self.raw_fastqs_keys = raw_fastqs_keys
         if self.metadata:
             bucket = self.metadata["workflowRoot"].split("gs://")[1].split("/")[0]
             if backend is None:
@@ -130,10 +133,11 @@ def get_files(self, filekey=None, filename=None):
     @property
     def raw_fastqs(self):
         fastqs = []
+        raw_fastqs_keys = ("fastq", "fastqs")
+        if self.raw_fastqs_keys is not None:
+            raw_fastqs_keys = self.raw_fastqs_keys
         for file in self.files:
-            if (
-                "fastqs" in file.filekeys or "fastq" in file.filekeys
-            ) and file.task is None:
+            if any([k in file.filekeys for k in raw_fastqs_keys]) and file.task is None:
                 fastqs.append(file)
         return fastqs
 

accession_steps/mirna_steps.json

@@ -7,7 +7,7 @@
         {
           "derived_from_files": [
             {
-              "derived_from_filekey": "fastq",
+              "derived_from_filekey": "fastqs_to_trim",
               "derived_from_inputs": "true",
               "derived_from_task": "cutadapt"
             },
@@ -33,7 +33,7 @@
         {
           "derived_from_files": [
             {
-              "derived_from_filekey": "fastq",
+              "derived_from_filekey": "fastqs_to_trim",
               "derived_from_inputs": "true",
               "derived_from_task": "cutadapt"
             },
@@ -110,5 +110,8 @@
       ],
       "wdl_task_name": "wigtobigwig"
     }
+  ],
+  "raw_fastqs_keys": [
+    "fastqs_to_trim"
   ]
 }