Add Kinnex alignment MetaWorkflowRun support

CHANGELOG.rst

@@ -3,13 +3,19 @@
 Change Log
 ==========
 
+3.12.0
+=====
+* Add Kinnex alignment MetaWorkflowRun creation function for SMaHT
+
+
 3.11.0
 =====
 * Most MetaWorkflowRun creation functions now accept multiple files or filesets as input.
 * Added `archive_broad_crams_and_fastqs` function to `wrangler_utils` to archive Broad submissions more easily. 
 * Added `purge_fileset` function to `wrangler_utils` to delete a fileset and its associated files.
 * Added `remove_property` function to `wrangler_utils` to remove properties from items.
 
+
 3.10.0
 =====
 * Added functionality to rerun a MetaWorkflowRun with imported steps from a previous run

magma_smaht/commands/create_meta_workflow_run.py

@@ -3,6 +3,7 @@
 from magma_smaht.create_metawfr import (
     mwfr_illumina_alignment,
     mwfr_rnaseq_alignment,
+    mwfr_kinnex_alignment,
     mwfr_pacbio_alignment,
     mwfr_fastqc,
     mwfr_hic_alignment,
@@ -91,6 +92,31 @@ def align_rnaseq(fileset_accessions, sequence_length, auth_env):
         mwfr_rnaseq_alignment(fileset_accession, sequence_length, smaht_key)
 
 
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-f",
+    "--fileset-accessions",
+    required=True,
+    multiple=True,
+    type=str,
+    help="Fileset accessions",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def align_kinnex(fileset_accessions, auth_env):
+    """Alignment MWFR for Kinnex data"""
+    smaht_key = get_auth_key(auth_env)
+    for fileset_accession in fileset_accessions:
+        print(f"Working on Fileset {fileset_accession}")
+        mwfr_kinnex_alignment(fileset_accession, smaht_key)
+
+
 @cli.command()
 @click.help_option("--help", "-h")
 @click.option(

magma_smaht/constants.py

@@ -5,9 +5,10 @@
 # MetaWorkflow names are used to get the latest version.
 # We assume that they don't change!
 MWF_NAME_ILLUMINA = "Illumina_alignment_GRCh38"
-MWF_NAME_RNASEQ = "RNA-seq_bulk_short_reads_GRCh38"
 MWF_NAME_ONT = "ONT_alignment_GRCh38"
 MWF_NAME_PACBIO = "PacBio_alignment_GRCh38"
+MWF_NAME_RNASEQ = "RNA-seq_bulk_short_reads_GRCh38"
+MWF_NAME_KINNEX = "RNA-seq_kinnex_long_reads_GRCh38"
 MWF_NAME_HIC = "Hi-C_alignment_GRCh38"
 MWF_NAME_FASTQC = "Illumina_FASTQ_quality_metrics"
 MWF_NAME_FASTQ_LONG_READ = "long_reads_FASTQ_quality_metrics"

magma_smaht/create_metawfr.py

@@ -36,6 +36,7 @@
 from magma_smaht.constants import (
     MWF_NAME_ILLUMINA,
     MWF_NAME_RNASEQ,
+    MWF_NAME_KINNEX,
     MWF_NAME_ONT,
     MWF_NAME_PACBIO,
     MWF_NAME_HIC,
@@ -145,6 +146,17 @@ def mwfr_rnaseq_alignment(fileset_accession, sequence_length, smaht_key):
     )
 
 
+def mwfr_kinnex_alignment(fileset_accession, smaht_key):
+    """Creates a MetaWorflowRun item in the portal for the Kinnex pipeline given a fileset"""
+
+    mwf = get_latest_mwf(MWF_NAME_KINNEX, smaht_key)
+    print(f"Using MetaWorkflow {mwf[ACCESSION]} ({mwf[ALIASES][0]})")
+
+    file_set = get_file_set(fileset_accession, smaht_key)
+    mwfr_input = get_core_alignment_mwfr_input(file_set, INPUT_FILES_BAM, smaht_key)
+    create_and_post_mwfr(mwf[UUID], file_set, INPUT_FILES_BAM, mwfr_input, smaht_key)
+
+
 def mwfr_pacbio_alignment(fileset_accession, smaht_key):
     """Creates a MetaWorflowRun item in the portal for PacBio alignment of submitted files within a fileset"""
 

magma_smaht/utils.py

@@ -326,6 +326,7 @@ def get_latest_mwf(mwf_name, smaht_key):
     Returns:
         dict: MWF item from portal
     """
+    #query = f"/search/?type=MetaWorkflow&version=0.3.0&name={mwf_name}"
     query = f"/search/?type=MetaWorkflow&name={mwf_name}"
     search_results = ff_utils.search_metadata(query, key=smaht_key)
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "magma-suite"
-version = "3.11.0"
+version = "3.12.0"
 description = "Collection of tools to manage meta-workflows automation."
 authors = ["Michele Berselli <[email protected]>", "Doug Rioux", "Soo Lee", "CGAP team"]
 license = "MIT"