Support SMaHT SNV/SV calling pipelines

magma_smaht/commands/create_meta_workflow_run.py

@@ -19,6 +19,13 @@
     mwfr_sample_identity_check,
     mwfr_bam_to_cram,
 )
+from magma_smaht.create_metawfr_variant_calling import (
+    mwfrs_somatic_snv_callers,
+    mwfrs_somatic_snv_callers_by_core,
+    mwfr_germline_snv_caller,
+    mwfr_somatic_snv_filtering,
+    mwfrs_somatic_sv_callers,
+)
 from magma_smaht.utils import get_auth_key
 
 
@@ -452,7 +459,6 @@ def conversion_bam_to_cram(fileset_accessions, auth_env):
     """Conversion MWFR for final output BAMs to CRAM"""
     smaht_key = get_auth_key(auth_env)
     mwfr_bam_to_cram(fileset_accessions, smaht_key)
-
 
 
 @cli.command()
@@ -461,7 +467,7 @@ def conversion_bam_to_cram(fileset_accessions, auth_env):
     "-f", "--files", required=True, type=str, multiple=True, help="BAM file accessions"
 )
 @click.option(
-    "-d", "--donor", required=True, type=str, help="Accession of the associated donor"
+    "-d", "--donor", required=False, type=str, help="Accession of the associated donor"
 )
 @click.option(
     "-e",
@@ -476,5 +482,142 @@ def sample_identity_check(files, donor, auth_env):
     mwfr_sample_identity_check(files, donor, smaht_key)
 
 
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option("-d", "--donor", required=True, type=str, help="Accession of the donor")
+@click.option(
+    "-t",
+    "--tissue",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of the donor-specific tissue used for variant calling (optional)",
+)
+@click.option(
+    "-a",
+    "--analysis-run",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of an existing analysis run (optional)",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def call_germline_snv(donor, tissue, analysis_run, auth_env):
+    """Call germline variants for a given donor. This is a prerequesite for somatic SNV calling."""
+    smaht_key = get_auth_key(auth_env)
+    mwfr_germline_snv_caller(donor, tissue, analysis_run, smaht_key)
+
+
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-t", "--tissue", required=True, type=str, help="Accession of donor specific tissue"
+)
+@click.option(
+    "-a",
+    "--analysis-run",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of an existing analysis run (optional)",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def call_somatic_snv_by_core(tissue, analysis_run, auth_env):
+    """Call SNV on a given tissue sample. Create the individual caller MWFRs by core. If an analysis run is provided, the MWFRs will be added to it."""
+    smaht_key = get_auth_key(auth_env)
+    mwfrs_somatic_snv_callers_by_core(tissue, analysis_run, smaht_key)
+
+
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-t", "--tissue", required=True, type=str, help="Accession of donor specific tissue"
+)
+@click.option(
+    "-a",
+    "--analysis-run",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of an existing analysis run (optional)",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def call_somatic_snv_step1(tissue, analysis_run, auth_env):
+    """Call SNV on a given tissue sample. Step 1 will create the individual caller MWFRs. If an analysis run is provided, the MWFRs will be added to it."""
+    smaht_key = get_auth_key(auth_env)
+    mwfrs_somatic_snv_callers(tissue, analysis_run, smaht_key)
+
+
+
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-t", "--tissue", required=True, type=str, help="Accession of donor specific tissue"
+)
+@click.option(
+    "-a",
+    "--analysis-run",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of an existing analysis run (optional)",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def call_somatic_sv(tissue, analysis_run, auth_env):
+    """Call somatic SV on a given tissue sample. This will create individual caller MWFRs. If an analysis run is provided, the MWFRs will be added to it."""
+    smaht_key = get_auth_key(auth_env)
+    mwfrs_somatic_sv_callers(tissue, analysis_run, smaht_key)
+
+
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-t", "--tissue", required=True, type=str, help="Accession of donor specific tissue"
+)
+@click.option(
+    "-a",
+    "--analysis-run",
+    required=False,
+    default=None,
+    type=str,
+    help="Accession of an existing analysis run",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def call_somatic_snv_step2(tissue, analysis_run, auth_env):
+    """Create filtered SNV on a given tissue sample. It will use the outputs of the caller MWFRs created in step 1. This step requires an existing analysis run with the caller MWFRs already added."""
+    smaht_key = get_auth_key(auth_env)
+    mwfr_somatic_snv_filtering(tissue, analysis_run, smaht_key)
+
+
 if __name__ == "__main__":
     cli()

magma_smaht/commands/wrangler_utils.py

@@ -10,6 +10,30 @@ def cli():
     pass
 
 
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-d",
+    "--donor",
+    required=True,
+    type=str,
+    help="External ID of the donor (e.g. SMHT001)",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def analysis_overview(donor, auth_env):
+    """
+    Check which files need to be checked for sample identity.
+    """
+    smaht_key = get_auth_key(auth_env)
+    wrangler_utils.analysis_overview(donor, smaht_key)
+
+
 @cli.command()
 @click.help_option("--help", "-h")
 @click.option(
@@ -78,6 +102,45 @@ def reset_mwfrs(mwfr_uuids, auth_env):
     wrangler_utils.reset_mwfrs(mwfr_uuids, smaht_key)
 
 
+@cli.command()
+@click.help_option("--help", "-h")
+@click.option(
+    "-m",
+    "--mwfr-uuids",
+    required=True,
+    type=str,
+    multiple=True,
+    help="List of MWFRs to reset",
+)
+@click.option(
+    "-p",
+    "--steps",
+    required=True,
+    type=str,
+    help="Steps to reset (comma-separated)",
+)
+@click.option(
+    "-s",
+    "--status",
+    required=True,
+    type=str,
+    help="Status to reset to",
+)
+@click.option(
+    "-e",
+    "--auth-env",
+    required=True,
+    type=str,
+    help="Name of environment in smaht-keys file",
+)
+def reset_status_mwfrs(mwfr_uuids, steps, status, auth_env):
+    """Reset a list of failed MetaWorkflowRuns to a desired status"""
+    smaht_key = get_auth_key(auth_env)
+    steps_list = steps.split(",")
+    for mwfr_uuid in mwfr_uuids:
+        wrangler_utils.reset_status_mwfr(mwfr_uuid, steps_list, status, smaht_key)
+
+
 @cli.command()
 @click.help_option("--help", "-h")
 @click.option(
@@ -87,17 +150,25 @@ def reset_mwfrs(mwfr_uuids, auth_env):
     type=str,
     help="Name of environment in smaht-keys file",
 )
+@click.option(
+    "-l",
+    "--limit",
+    required=False,
+    default=100,
+    type=int,
+    help="Limit of failed MWFRs to reset",
+)
 @click.option(
     "--ignore-md5",
     default=False,
     is_flag=True,
     show_default=True,
     help="Ignore MD5 checksum runs",
 )
-def reset_all_failed_mwfrs(auth_env, ignore_md5):
+def reset_all_failed_mwfrs(auth_env, ignore_md5, limit):
     """Reset all failed MetaWorkflowRuns on the portal"""
     smaht_key = get_auth_key(auth_env)
-    wrangler_utils.reset_all_failed_mwfrs(smaht_key, ignore_md5)
+    wrangler_utils.reset_all_failed_mwfrs(smaht_key, ignore_md5, limit)
 
 
 @cli.command()

magma_smaht/constants.py

@@ -20,6 +20,20 @@
 MWF_NAME_ULTRA_LONG_BAMQC = "ultra-long_reads_BAM_quality_metrics_GRCh38"
 MWF_NAME_LONG_READ_BAMQC = "long_reads_BAM_quality_metrics_GRCh38"
 MWF_SAMPLE_IDENTITY_CHECK = "sample_identity_check"
+# Variant callering
+MWF_NAME_TNHAPLOTYPER = "paired-end_short_reads_variant_calling_TNhaplotyper2_distributed_TNfilter_GRCh38"
+MWF_NAME_LONGCALLD = "PacBio_variant_calling_longcallD_GRCh38"
+MWF_NAME_LONGCALLD_SINGLE_FILE = "PacBio_variant_calling_longcallD_single_file_GRCh38"
+MWF_NAME_STRELKA2 = "paired-end_short_reads_variant_calling_Strelka2_distributed_GRCh38"
+MWF_NAME_RUFUS = "paired-end_short_reads_variant_calling_RUFUS_distributed_GRCh38"
+MWF_NAME_DNASCOPEHYBRID = "paired-end_short_plus_PacBio_variant_calling_DNAscopeHybrid_GRCh38"
+MWF_NAME_SNIFFLES = "PacBio_ONT_variant_calling_Sniffles_GRCh38"
+MWF_NAME_SEVERUS = "PacBio_ONT_variant_calling_Severus_GRCh38"
+MWF_NAME_DELLY = "PacBio_ONT_variant_calling_Delly_GRCh38"
+MWF_NAME_SNV_FILTERING_LONGCALLD = "SNV_filtering_longcallD_GRCh38"
+MWF_NAME_SNV_FILTERING = "SNV_filtering_GRCh38"
+
+HMS_DAC_UUID = "9626d82e-8110-4213-ac75-0a50adf890ff"
 
 # Input argument names
 INPUT_FILES_R1_FASTQ_GZ = "input_files_r1_fastq_gz"
@@ -28,6 +42,8 @@
 INPUT_FILES = "input_files"
 INPUT_FILES_FASTQ_GZ = "input_files_fastq_gz"
 INPUT_FILES_CRAM = "input_files_cram"
+INPUT_FILE_CRAM = "input_file_cram"
+ADDITIONAL_INPUT_FILES_CRAM = "additional_input_files_cram"
 GENOME_REFERENCE_FASTA = "genome_reference_fasta"
 SAMPLE_NAME = "sample_name"
 SAMPLE_NAMES = "sample_names"
@@ -46,7 +62,9 @@
 SEQUENCING_CENTER = "sequencing_center"
 CONSORTIA = "consortia"
 FILE_SETS = "file_sets"
-META_WORFLOW_RUN = "MetaWorkflowRun"
+ANALYSIS_RUNS = "analysis_runs"
+META_WORKFLOW_RUN = "MetaWorkflowRun"
+ANALYSIS_RUN = "AnalysisRun"
 ACCESSION = "accession"
 DISPLAY_TITLE = "display_title"
 ALIASES = "aliases"
@@ -63,6 +81,10 @@
 FIRST_STRANDED = "First Stranded"
 SECOND_STRANDED = "Second Stranded"
 FAILED_JOBS = "failed_jobs"
+SOMATIC_SNV_CALLING = "Somatic SNV calling"
+SOMATIC_SNV_CALLING_CORE_SPECIFIC = "Somatic SNV calling (core specific)"
+SOMATIC_SV_CALLING = "Somatic SV calling"
+GERMLINE_SNV_CALLING = "Germline SNV calling"
 
 # Assays
 WGS = "WGS"