Add nallo upload

cg/cli/upload/base.py

@@ -27,8 +27,8 @@
     upload_rna_junctions_to_scout,
     upload_rna_omics_to_scout,
     upload_rna_to_scout,
-    upload_tomte_to_scout,
     upload_to_scout,
+    upload_tomte_to_scout,
 )
 from cg.cli.upload.utils import suggest_cases_to_upload
 from cg.cli.upload.validate import validate
@@ -40,9 +40,10 @@
 from cg.meta.upload.mip.mip_dna import MipDNAUploadAPI
 from cg.meta.upload.mip.mip_rna import MipRNAUploadAPI
 from cg.meta.upload.mutant.mutant import MutantUploadAPI
+from cg.meta.upload.nallo.nallo import NalloUploadAPI
 from cg.meta.upload.nf_analysis import NfAnalysisUploadAPI
-from cg.meta.upload.tomte.tomte import TomteUploadAPI
 from cg.meta.upload.raredisease.raredisease import RarediseaseUploadAPI
+from cg.meta.upload.tomte.tomte import TomteUploadAPI
 from cg.meta.upload.upload_api import UploadAPI
 from cg.models.cg_config import CGConfig
 from cg.store.models import Case
@@ -86,6 +87,8 @@ def upload(context: click.Context, case_id: str | None, restart: bool):
             upload_api = MipRNAUploadAPI(config_object)
         elif case.data_analysis == Workflow.MICROSALT:
             upload_api = MicrosaltUploadAPI(config_object)
+        elif case.data_analysis == Workflow.NALLO:
+            upload_api = NalloUploadAPI(config_object)
         elif case.data_analysis == Workflow.RAREDISEASE:
             upload_api = RarediseaseUploadAPI(config_object)
         elif case.data_analysis == Workflow.TOMTE:

cg/cli/upload/scout.py

@@ -20,6 +20,7 @@
 from cg.meta.workflow.balsamic_umi import BalsamicUmiAnalysisAPI
 from cg.meta.workflow.mip_dna import MipDNAAnalysisAPI
 from cg.meta.workflow.mip_rna import MipRNAAnalysisAPI
+from cg.meta.workflow.nallo import NalloAnalysisAPI
 from cg.meta.workflow.raredisease import RarediseaseAnalysisAPI
 from cg.meta.workflow.rnafusion import RnafusionAnalysisAPI
 from cg.meta.workflow.tomte import TomteAnalysisAPI
@@ -316,6 +317,7 @@ def get_upload_api(case: Case, cg_config: CGConfig) -> UploadAPI:
         Workflow.BALSAMIC_UMI: BalsamicUmiAnalysisAPI,
         Workflow.MIP_RNA: MipRNAAnalysisAPI,
         Workflow.MIP_DNA: MipDNAAnalysisAPI,
+        Workflow.NALLO: NalloAnalysisAPI,
         Workflow.RAREDISEASE: RarediseaseAnalysisAPI,
         Workflow.RNAFUSION: RnafusionAnalysisAPI,
         Workflow.TOMTE: TomteAnalysisAPI,

cg/constants/housekeeper_tags.py

@@ -105,6 +105,11 @@ class BalsamicAnalysisTag:
     QC_METRICS: list[str] = ["qc-metrics", "deliverable"]
 
 
+class NalloAnalysisTag:
+    HAPLOTAGS: str = "haplotags"
+    PARAPHASE: str = "paraphase"
+
+
 class HkAnalysisMetricsTag:
     QC_METRICS: set[str] = {"qc-metrics", "deliverable"}
 

cg/constants/scout.py

@@ -1,7 +1,7 @@
 from enum import StrEnum, auto
 
 from cg.constants import FileExtensions
-from cg.constants.housekeeper_tags import AlignmentFileTag
+from cg.constants.housekeeper_tags import AlignmentFileTag, NalloAnalysisTag
 
 HGNC_ID = "hgnc_id"
 
@@ -79,6 +79,18 @@ class ScoutUploadKey(StrEnum):
     vcf_str={"vcf-str"},
 )
 
+NALLO_CASE_TAGS = dict(
+    delivery_report={"delivery-report"},
+    multiqc={"multiqc-html"},
+    somalier_pairs={"relate-pairs", "somalier"},
+    somalier_samples={"relate-samples", "somalier"},
+    vcf_snv_research={"vcf-snv-research"},
+    vcf_snv={"vcf-snv-clinical"},
+    vcf_sv_research={"vcf-sv-research"},
+    vcf_sv={"vcf-sv-clinical"},
+    vcf_str={"vcf-str"},
+)
+
 BALSAMIC_CASE_TAGS = dict(
     sv_vcf={"vcf-sv-clinical"},
     snv_vcf={"vcf-snv-clinical"},
@@ -138,6 +150,12 @@ class ScoutUploadKey(StrEnum):
     mitodel_file={"mitodel"},
 )
 
+NALLO_SAMPLE_TAGS: dict[str, set[str]] = dict(
+    alignment_file={AlignmentFileTag.BAM, NalloAnalysisTag.HAPLOTAGS},
+    d4_file={"d4"},
+    paraphase_alignment_path={AlignmentFileTag.BAM, NalloAnalysisTag.PARAPHASE},
+)
+
 BALSAMIC_SAMPLE_TAGS = dict(
     bam_file={"bam"},
     alignment_file={"cram"},
@@ -155,3 +173,7 @@ class ScoutUploadKey(StrEnum):
 )
 
 RANK_MODEL_THRESHOLD = 5
+
+NALLO_RANK_MODEL_THRESHOLD = 8
+NALLO_RANK_MODEL_VERSION_SNV = "1.0"
+NALLO_RANK_MODEL_VERSION_SV = "1.0"

cg/meta/upload/nallo/nallo.py

@@ -0,0 +1,43 @@
+"""NALLO upload API."""
+
+import datetime as dt
+import logging
+
+import rich_click as click
+
+from cg.cli.generate.delivery_report.base import generate_delivery_report
+from cg.cli.upload.scout import upload_to_scout
+from cg.constants import REPORT_SUPPORTED_DATA_DELIVERY, DataDelivery
+from cg.meta.upload.upload_api import UploadAPI
+from cg.meta.workflow.nallo import NalloAnalysisAPI
+from cg.models.cg_config import CGConfig
+from cg.store.models import Analysis, Case
+
+LOG = logging.getLogger(__name__)
+
+
+class NalloUploadAPI(UploadAPI):
+    """NALLO upload API."""
+
+    def __init__(self, config: CGConfig):
+        self.analysis_api = NalloAnalysisAPI(config)
+        super().__init__(config=config, analysis_api=self.analysis_api)
+
+    def upload(self, ctx: click.Context, case: Case, restart: bool) -> None:
+        """Uploads NALLO analysis data and files."""
+        analysis: Analysis = case.analyses[0]
+        self.update_upload_started_at(analysis=analysis)
+        # Delivery report generation
+        if case.data_delivery in REPORT_SUPPORTED_DATA_DELIVERY:
+            ctx.invoke(generate_delivery_report, case_id=case.internal_id)
+
+        # Scout specific upload
+        if DataDelivery.SCOUT in case.data_delivery:
+            ctx.invoke(upload_to_scout, case_id=case.internal_id, re_upload=restart)
+        LOG.info(
+            f"Upload of case {case.internal_id} was successful. Uploaded at {dt.datetime.now()} in StatusDB"
+        )
+        self.update_uploaded_at(analysis=analysis)
+
+        # Clinical delivery upload
+        self.upload_files_to_customer_inbox(case)

cg/meta/upload/scout/hk_tags.py

@@ -32,6 +32,9 @@ class CaseTags(BaseModel):
     peddy_ped: set[str] = Field(None, description="Ped info from peddy, only for rare disease")
     peddy_sex: set[str] | None = Field(None, description="Peddy sex check, only for rare disease")
     peddy_check: set[str] = Field(None, description="Peddy pedigree check, only for rare disease")
+    somalier_samples: set[str] = Field(None, description="Somalier samples info")
+    somalier_pairs: set[str] = Field(None, description="Somalier pairs info")
+    sex_check: set[str] = Field(None, description="Somalier sex check info, only for rare disease")
     multiqc_report: set[str] | None = Field(None, description="MultiQC report")
     multiqc: set[str] | None = Field(None, description="MultiQC report")
     delivery_report: set[str] | None = Field(None, description="Delivery report")
@@ -84,3 +87,4 @@ class SampleTags(BaseModel):
     reviewer_catalog: set[str] | None = None
     reviewer_vcf: set[str] | None = None
     mitodel_file: set[str] | None = None
+    paraphase_alignment_path: set[str] | None = None

cg/meta/upload/scout/nallo_config_builder.py

@@ -0,0 +1,88 @@
+import logging
+
+from housekeeper.store.models import Version
+
+from cg.apps.lims import LimsAPI
+from cg.apps.madeline.api import MadelineAPI
+from cg.constants.housekeeper_tags import HK_DELIVERY_REPORT_TAG
+from cg.constants.scout import (
+    NALLO_CASE_TAGS,
+    NALLO_RANK_MODEL_THRESHOLD,
+    NALLO_RANK_MODEL_VERSION_SNV,
+    NALLO_RANK_MODEL_VERSION_SV,
+    NALLO_SAMPLE_TAGS,
+    GenomeBuild,
+    UploadTrack,
+)
+from cg.meta.upload.scout.hk_tags import CaseTags, SampleTags
+from cg.meta.upload.scout.scout_config_builder import ScoutConfigBuilder
+from cg.meta.workflow.nallo import NalloAnalysisAPI
+from cg.models.scout.scout_load_config import NalloLoadConfig, ScoutNalloIndividual
+from cg.store.models import Analysis
+
+LOG = logging.getLogger(__name__)
+
+
+class NalloConfigBuilder(ScoutConfigBuilder):
+    def __init__(
+        self,
+        hk_version_obj: Version,
+        analysis_obj: Analysis,
+        nallo_analysis_api: NalloAnalysisAPI,
+        lims_api: LimsAPI,
+        madeline_api: MadelineAPI,
+    ):
+        super().__init__(
+            hk_version_obj=hk_version_obj,
+            analysis_obj=analysis_obj,
+            lims_api=lims_api,
+        )
+        self.case_tags: CaseTags = CaseTags(**NALLO_CASE_TAGS)
+        self.sample_tags: SampleTags = SampleTags(**NALLO_SAMPLE_TAGS)
+        self.nallo_analysis_api: NalloAnalysisAPI = nallo_analysis_api
+        self.lims_api: LimsAPI = lims_api
+        self.madeline_api: MadelineAPI = madeline_api
+
+    def build_load_config(self) -> NalloLoadConfig:
+        """Create a NALLO specific load config for uploading analysis to Scout."""
+        LOG.info("Build load config for NALLO case")
+        load_config: NalloLoadConfig = NalloLoadConfig(
+            track=UploadTrack.RARE_DISEASE.value,
+            delivery_report=self.get_file_from_hk({HK_DELIVERY_REPORT_TAG}),
+        )
+        self.add_common_info_to_load_config(load_config)
+        load_config.gene_panels = self.nallo_analysis_api.get_aggregated_panels(
+            customer_id=self.analysis_obj.case.customer.internal_id,
+            default_panels=set(self.analysis_obj.case.panels),
+        )
+        self.include_case_files(load_config)
+        self.get_sample_information(load_config)
+        self.include_pedigree_picture(load_config)
+        load_config.human_genome_build = GenomeBuild.hg38
+        load_config.rank_score_threshold = NALLO_RANK_MODEL_THRESHOLD
+        load_config.rank_model_version = NALLO_RANK_MODEL_VERSION_SNV
+        load_config.sv_rank_model_version = NALLO_RANK_MODEL_VERSION_SV
+        return load_config
+
+    def include_case_files(self, load_config: NalloLoadConfig) -> None:
+        """Include case level files for NALLO case."""
+        LOG.info("Including NALLO specific case level files")
+        for scout_key in NALLO_CASE_TAGS.keys():
+            self._include_case_file(load_config, scout_key)
+
+    def _include_case_file(self, load_config: NalloLoadConfig, scout_key: str) -> None:
+        """Include the file path associated to a scout configuration parameter if the corresponding housekeeper tags
+        are found. Otherwise return None."""
+        file_path = self.get_file_from_hk(getattr(self.case_tags, scout_key))
+        setattr(load_config, scout_key, file_path)
+
+    def include_sample_files(self, config_sample: ScoutNalloIndividual) -> None:
+        """Include sample level files that are optional."""
+        LOG.info("Including NALLO specific sample level files")
+        sample_id: str = config_sample.sample_id
+        config_sample.d4_file = self.get_sample_file(
+            hk_tags=self.sample_tags.d4_file, sample_id=sample_id
+        )
+        config_sample.paraphase_alignment_path = self.get_sample_file(
+            hk_tags=self.sample_tags.paraphase_alignment_path, sample_id=sample_id
+        )

cg/meta/upload/scout/raredisease_config_builder.py

@@ -148,7 +148,7 @@ def load_custom_image_sample(self, load_config: RarediseaseLoadConfig) -> None:
             load_config.custom_images = config_custom_images
 
     def include_case_files(self, load_config: RarediseaseLoadConfig) -> None:
-        """Include case level files for mip case."""
+        """Include case level files for RAREDISEASE case."""
         LOG.info("Including RAREDISEASE specific case level files")
         for scout_key in RAREDISEASE_CASE_TAGS.keys():
             self._include_case_file(load_config, scout_key)

cg/meta/upload/scout/scout_config_builder.py

@@ -16,6 +16,7 @@
     ScoutIndividual,
     ScoutLoadConfig,
     ScoutMipIndividual,
+    ScoutNalloIndividual,
     ScoutRarediseaseIndividual,
 )
 from cg.store.models import Analysis, Case, CaseSample, Sample
@@ -107,6 +108,8 @@ def build_config_sample(self, case_sample: CaseSample) -> ScoutIndividual:
             config_sample = ScoutRarediseaseIndividual()
         elif self.analysis_obj.workflow == Workflow.MIP_DNA:
             config_sample = ScoutMipIndividual()
+        elif self.analysis_obj.workflow == Workflow.NALLO:
+            config_sample = ScoutNalloIndividual()
         elif self.analysis_obj.workflow == Workflow.RNAFUSION:
             config_sample = ScoutIndividual()
         self.add_common_sample_info(config_sample=config_sample, case_sample=case_sample)
@@ -158,7 +161,7 @@ def include_sample_files(self, config_sample: ScoutIndividual) -> None:
         """Include all files that are used on sample level in Scout."""
         raise NotImplementedError
 
-    def include_case_files(self) -> None:
+    def include_case_files(self, load_config: ScoutLoadConfig) -> None:
         """Include all files that are used on case level in Scout."""
         raise NotImplementedError
 

cg/meta/upload/scout/uploadscoutapi.py

@@ -19,6 +19,7 @@
 from cg.meta.upload.scout.balsamic_config_builder import BalsamicConfigBuilder
 from cg.meta.upload.scout.balsamic_umi_config_builder import BalsamicUmiConfigBuilder
 from cg.meta.upload.scout.mip_config_builder import MipConfigBuilder
+from cg.meta.upload.scout.nallo_config_builder import NalloConfigBuilder
 from cg.meta.upload.scout.raredisease_config_builder import RarediseaseConfigBuilder
 from cg.meta.upload.scout.rnafusion_config_builder import RnafusionConfigBuilder
 from cg.meta.upload.scout.scout_config_builder import ScoutConfigBuilder
@@ -632,6 +633,13 @@ def get_config_builder(self, analysis, hk_version) -> ScoutConfigBuilder:
                 lims_api=self.lims,
                 madeline_api=self.madeline_api,
             ),
+            Workflow.NALLO: NalloConfigBuilder(
+                hk_version_obj=hk_version,
+                analysis_obj=analysis,
+                nallo_analysis_api=self.analysis_api,
+                lims_api=self.lims,
+                madeline_api=self.madeline_api,
+            ),
             Workflow.RAREDISEASE: RarediseaseConfigBuilder(
                 hk_version_obj=hk_version,
                 analysis_obj=analysis,

cg/models/scout/scout_load_config.py

@@ -82,6 +82,11 @@ class ScoutMipIndividual(ScoutIndividual):
     mitodel_file: str | None = None
 
 
+class ScoutNalloIndividual(ScoutIndividual):
+    d4_file: str | None = None
+    paraphase_alignment_path: str | None = None
+
+
 class ScoutRarediseaseIndividual(ScoutIndividual):
     mt_bam: str | None = None
     chromograph_images: ChromographImages = ChromographImages()
@@ -163,6 +168,18 @@ class MipLoadConfig(ScoutLoadConfig):
     vcf_sv_research: Annotated[str | None, BeforeValidator(field_not_none)] = None
 
 
+class NalloLoadConfig(ScoutLoadConfig):
+    madeline: str | None = None
+    samples: list[ScoutNalloIndividual] = []
+    somalier_samples: str | None = None
+    somalier_pairs: str | None = None
+    vcf_snv: Annotated[str, BeforeValidator(field_not_none)] = None
+    vcf_snv_research: Annotated[str | None, BeforeValidator(field_not_none)] = None
+    vcf_sv: Annotated[str | None, BeforeValidator(field_not_none)] = None
+    vcf_sv_research: Annotated[str | None, BeforeValidator(field_not_none)] = None
+    vcf_str: str | None = None
+
+
 class RarediseaseLoadConfig(ScoutLoadConfig):
     madeline: str | None = None
     peddy_check: str | None = None

tests/cli/upload/test_cli_scout.py

@@ -12,6 +12,7 @@
 from cg.meta.workflow.balsamic import BalsamicAnalysisAPI
 from cg.meta.workflow.mip_dna import MipDNAAnalysisAPI
 from cg.meta.workflow.mip_rna import MipRNAAnalysisAPI
+from cg.meta.workflow.nallo import NalloAnalysisAPI
 from cg.meta.workflow.rnafusion import RnafusionAnalysisAPI
 from cg.meta.workflow.tomte import TomteAnalysisAPI
 from cg.models.cg_config import CGConfig
@@ -24,13 +25,15 @@
     Workflow.BALSAMIC,
     Workflow.MIP_DNA,
     Workflow.MIP_RNA,
+    Workflow.NALLO,
     Workflow.RNAFUSION,
 ]
 
 ANALYSIS_API: list = [
     (Workflow.BALSAMIC, BalsamicAnalysisAPI),
     (Workflow.MIP_DNA, MipDNAAnalysisAPI),
     (Workflow.MIP_RNA, MipRNAAnalysisAPI),
+    (Workflow.NALLO, NalloAnalysisAPI),
     (Workflow.RNAFUSION, RnafusionAnalysisAPI),
     (Workflow.TOMTE, TomteAnalysisAPI),
 ]

tests/cli/upload/test_cli_upload_auto.py

@@ -2,8 +2,8 @@
 
 import datetime
 import logging
-import pytest
 
+import pytest
 from click.testing import CliRunner
 
 from cg.cli.upload.base import upload_all_completed_analyses
@@ -16,6 +16,7 @@
     Workflow.MICROSALT,
     Workflow.MIP_DNA,
     Workflow.MIP_RNA,
+    Workflow.NALLO,
     Workflow.RAREDISEASE,
     Workflow.RNAFUSION,
     Workflow.TAXPROFILER,