feat: add support for SV TSV files (#75) (#78)

clinvar_this/batches.py

@@ -122,18 +122,26 @@ def import_(config: config.Config, name: str, path: str, metadata: typing.Tuple[
         logger.info("Creating new payload only")
         previous_submission_container = None
     if path.endswith(".tsv") or path.endswith(".txt"):
-        tsv_records = tsv.read_tsv(path=path)
-        batch_metadata = tsv.batch_metadata_from_mapping(metadata, use_defaults=True)
-        new_submission_container = tsv.tsv_records_to_submission_container(
-            tsv_records, batch_metadata
-        )
-        if previous_submission_container:
-            submission_container = _merge_submission_container(
-                base=previous_submission_container,
-                patch=new_submission_container,
-            )
+        tsv_type = tsv.guess_tsv_type(path)
+        if tsv_type in (tsv.TsvType.SEQ_VAR, tsv.TsvType.STRUC_VAR):
+            batch_metadata = tsv.batch_metadata_from_mapping(metadata, use_defaults=True)
+            if tsv_type == tsv.TsvType.SEQ_VAR:
+                new_submission_container = tsv.seq_var_tsv_records_to_submission_container(
+                    tsv.read_seq_var_tsv(path=path), batch_metadata
+                )
+            else:  # tsv_type == tsv.TsvType.STRUC_VAR
+                new_submission_container = tsv.struc_var_tsv_records_to_submission_container(
+                    tsv.read_struc_var_tsv(path=path), batch_metadata
+                )
+            if previous_submission_container:
+                submission_container = _merge_submission_container(
+                    base=previous_submission_container,
+                    patch=new_submission_container,
+                )
+            else:
+                submission_container = new_submission_container
         else:
-            submission_container = new_submission_container
+            raise exceptions.IOException(f"Could not guess TSV file type from header for {path}")
         _write_payload(submission_container, config.profile, name)
     else:  # pragma: no cover
         raise exceptions.IOException(f"File extension of {path} cannot be handled.")
@@ -152,16 +160,24 @@ def _load_latest_payload(profile: str, name: str):
     return common.CONVERTER.structure(payload_unstructured, models.SubmissionContainer)
 
 
-def export(config: config.Config, name: str, path: str, force: bool = False):
+def export(
+    config: config.Config, name: str, path: str, force: bool = False, struc_var: bool = False
+):
     """Export the batch with the given ``name`` to the file at ``path``."""
     if pathlib.Path(path).exists() and not force:
         raise exceptions.IOException(
             f"File at output path {path} already exists. Use --force to overwrite."
         )
     if path.endswith(".tsv") or path.endswith(".txt"):
         payload = _load_latest_payload(config.profile, name)
-        tsv_records = tsv.submission_container_to_tsv_records(payload)
-        tsv.write_tsv(tsv_records, path=path)
+        if struc_var:
+            tsv.write_struc_var_tsv(
+                tsv_records=tsv.submission_container_to_struc_var_tsv_records(payload), path=path
+            )
+        else:
+            tsv.write_seq_var_tsv(
+                tsv_records=tsv.submission_container_to_seq_var_tsv_records(payload), path=path
+            )
     else:  # pragma: no cover
         raise exceptions.IOException(f"File extension of {path} cannot be handled.")
 

clinvar_this/cli.py

@@ -118,16 +118,23 @@ def batch_import(
 @click.argument("name")
 @click.argument("path")
 @click.option("--force/--no-force", required=False, default=False, help="Overwrite existing files")
+@click.option(
+    "--struc-var/--no-struc-var",
+    required=False,
+    default=False,
+    help="Export structural variants rather than sequence variants",
+)
 @click.pass_context
 def batch_export(
     ctx: click.Context,
     name: str,
     path: str,
     force: bool = False,
+    struc_var: bool = False,
 ):
     """Export batch data to a given file"""
     config_obj = load_config(ctx.obj["profile"])
-    batches.export(config_obj, name, path, force)
+    batches.export(config_obj, name, path, force, struc_var)
 
 
 @batch.command("update-metadata")