Sn DSA Sample Information

CHANGELOG.rst

@@ -7,6 +7,11 @@ smaht-portal
 Change Log
 ----------
 
+0.109.1
+=======
+`PR 286 Sn DSA Sample Information <https://github.com/smaht-dac/smaht-portal/pull/286>`_
+* Modify calculated properties for Supplementary File to show up in File Overview Page: `sample_summary` and `data_generation_summary` to pull from `DonorSpecificAssembly.derived_from.file_sets`
+
 0.109.0
 =======
 `PR 280 SN Assay info <https://github.com/smaht-dac/smaht-portal/pull/280>`_

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "encoded"
-version = "0.109.0"
+version = "0.109.1"
 description = "SMaHT Data Analysis Portal"
 authors = ["4DN-DCIC Team <[email protected]>"]
 license = "MIT"

src/encoded/item_utils/file.py

@@ -420,4 +420,9 @@ def get_associated_files_status(
             request_handler,
             get_file_sets(file),
             partial(file_set.get_associated_files_status, request_handler, at_id)
-    )
+    )
+
+
+def get_override_group_coverage(file: Dict[str, Any]) -> str:
+    """Get override group coverage from properties."""
+    return file.get("override_group_coverage","")

src/encoded/item_utils/supplementary_file.py

@@ -1,9 +1,19 @@
 from typing import Any, Dict, Union, Optional, List
+from functools import partial
+
 from .utils import RequestHandler, get_property_value_from_identifier, get_property_values_from_identifiers
 from . import (
     donor_specific_assembly as dsa_utils,
     file as file_utils,
-    item as item_utils
+    item as item_utils,
+    file_set as file_set_utils,
+    library as library_utils,
+    sample as sample_utils,
+    tissue as tissue_utils,
+    cell_culture as cell_culture_utils,
+    cell_culture_mixture as cell_culture_mixture_utils,
+    cell_line as cell_line_utils,
+    sequencing as sequencing_utils
 )
 
 def is_supplementary_file(properties: Dict[str, Any]) -> bool:
@@ -66,6 +76,68 @@ def get_derived_from_file_sets(
     return properties.get("file_sets", [])  
 
 
+def get_sequencings(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get sequencings associated with file."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler,
+            get_derived_from_file_sets(properties, request_handler), 
+            file_set_utils.get_sequencing
+        )
+    return properties.get("sequencing", [])
+
+
+def get_sequencers(
+    properties: Dict[str, Any], request_handler: RequestHandler
+) -> List[str]:
+    """Get sequencers associated with file."""
+    sequencings = get_sequencings(properties, request_handler)
+    return get_property_values_from_identifiers(
+        request_handler, sequencings, sequencing_utils.get_sequencer
+    )
+
+
+def get_libraries(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get libraries associated with file from derived_from files."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler, 
+            get_derived_from_file_sets(properties,request_handler),
+            file_set_utils.get_libraries
+        )
+    return properties.get("libraries", [])
+
+
+def get_analytes(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get analytes associated with file from derived_from files."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler,
+            get_libraries(properties, request_handler),
+            library_utils.get_analytes,
+        )
+    return properties.get("analytes", [])
+
+
+def get_assays(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get assays associated with file."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler,
+            get_libraries(properties, request_handler),
+            library_utils.get_assay,
+        )
+    return properties.get("assays", [])
+
+
 def get_dsa_software(
         properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None 
 ) -> List[Union[str, Dict[str, Any]]]:
@@ -76,4 +148,130 @@ def get_dsa_software(
             get_donor_specific_assembly(properties),
             dsa_utils.get_software,
         )
-    return []
+    return []
+
+
+def get_samples(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get samples associated with file."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler,
+            get_derived_from_file_sets(properties, request_handler),
+            partial(file_set_utils.get_samples, request_handler=request_handler),
+        )
+    return properties.get("samples", [])
+
+
+def get_sample_sources(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get sample sources associated with file."""
+    if request_handler:
+        return get_property_values_from_identifiers(
+            request_handler,
+            get_samples(properties, request_handler),
+            sample_utils.get_sample_sources,
+        )
+    return properties.get("sample_sources", [])
+
+
+def get_tissues(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get tissues associated with file."""
+    sample_sources = get_sample_sources(properties, request_handler=request_handler)
+    if request_handler:
+        return [
+            sample_source
+            for sample_source in sample_sources
+            if tissue_utils.is_tissue(request_handler.get_item(sample_source))
+        ]
+    return [
+        sample_source
+        for sample_source in sample_sources
+        if isinstance(sample_source, dict) and tissue_utils.is_tissue(sample_source)
+    ]
+
+
+def get_cell_culture_mixtures(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get cell culture mixtures associated with file."""
+    sample_sources = get_sample_sources(properties, request_handler=request_handler)
+    if request_handler:
+        return [
+            sample_source
+            for sample_source in sample_sources
+            if cell_culture_mixture_utils.is_cell_culture_mixture(
+                request_handler.get_item(sample_source)
+            )
+        ]
+    return [
+        sample_source
+        for sample_source in sample_sources
+        if isinstance(sample_source, dict)
+        and cell_culture_mixture_utils.is_cell_culture_mixture(sample_source)
+    ]
+
+
+def get_cell_cultures(
+    properties: Dict[str, Any], request_handler: RequestHandler
+) -> List[str]:
+    """Get cell cultures associated with file."""
+    sample_sources = get_sample_sources(properties, request_handler=request_handler)
+    cell_culture_mixtures = get_cell_culture_mixtures(
+        properties, request_handler=request_handler
+    )
+    cell_cultures_from_mixtures = get_property_values_from_identifiers(
+        request_handler, cell_culture_mixtures, cell_culture_mixture_utils.get_cell_cultures
+    )
+    direct_cell_cultures = [
+        sample_source
+        for sample_source in sample_sources
+        if cell_culture_utils.is_cell_culture(request_handler.get_item(sample_source))
+    ]
+    return list(set(cell_cultures_from_mixtures + direct_cell_cultures))
+
+
+def get_cell_lines(
+    properties: Dict[str, Any], request_handler: RequestHandler
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get cell lines associated with file."""
+    cell_cultures = get_cell_cultures(properties, request_handler)
+    cell_culture_mixtures = get_cell_culture_mixtures(
+        properties, request_handler=request_handler
+    )
+    return list(
+        set(
+            get_property_values_from_identifiers(
+                request_handler, cell_cultures, cell_culture_utils.get_cell_line
+            )
+            + get_property_values_from_identifiers(
+                request_handler,
+                cell_culture_mixtures,
+                partial(cell_culture_mixture_utils.get_cell_lines, request_handler),
+            )
+        )
+    )
+
+
+def get_donors(
+    properties: Dict[str, Any], request_handler: Optional[RequestHandler] = None
+) -> List[Union[str, Dict[str, Any]]]:
+    """Get donors associated with file."""
+    if request_handler:
+        tissues = get_tissues(properties, request_handler)
+        cell_lines = get_cell_lines(properties, request_handler)
+        return list(
+            set(
+                get_property_values_from_identifiers(
+                    request_handler, tissues, tissue_utils.get_donor
+                )
+                + get_property_values_from_identifiers(
+                    request_handler, cell_lines, partial(cell_line_utils.get_source_donor, request_handler)
+                )
+            )
+        )
+    return properties.get("donors", [])

src/encoded/schemas/donor_specific_assembly.json

@@ -23,6 +23,9 @@
         {
             "$ref": "mixins.json#/attribution"
         },
+        {
+            "$ref": "mixins.json#/code"
+        },
         {
             "$ref": "mixins.json#/description"
         },
@@ -64,6 +67,9 @@
         "version": {
             "description": "Version of the assembly"
         },
+        "code": {
+            "pattern": "^[A-Za-z0-9]{3,}$"
+        },
         "submitted_id": {
             "pattern": "^[A-Z0-9]{3,}_DONOR-SPECIFIC-ASSEMBLY_[A-Z0-9-_.]{4,}$"
         },