Added metadata column selection to spiec-easi and flashweave

BenKaehler · Sep 29, 2023 · 75b3eb6 · 75b3eb6
1 parent 715778a
commit 75b3eb6
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 2 deletions.
diff --git a/q2_makarsa/_flashweave.py b/q2_makarsa/_flashweave.py
@@ -12,6 +12,7 @@
 def flashweave(
     table: pd.DataFrame,
     metadata: qiime2.Metadata = None,
+    metadata_column: list = None,
     # defaults copied from FlashWeave.jl/src/learning.jl
     heterogeneous: bool = False,
     sensitive: bool = True,
@@ -64,6 +65,8 @@ def flashweave(
         ]
         if metadata:
             metadata = metadata.to_dataframe()
+            if metadata_column:
+                metadata = metadata[metadata_column]
             if set(table.index) <= set(metadata.index):
                 metadata = metadata.reindex(table.index)
                 metadata_file = temp_dir / "meta-input-data.tsv"

diff --git a/q2_makarsa/_spieceasi.py b/q2_makarsa/_spieceasi.py
@@ -11,9 +11,11 @@
 from ._run_commands import run_commands
 
 
-def encode_metadata(sample_ids, metadata):
+def encode_metadata(sample_ids, metadata, metadata_column):
     encoded_metadata = []
     for column in metadata.columns:
+        if metadata_column and column not in metadata_column:
+            continue
         values = metadata.get_column(column)
 
         is_numeric = isinstance(values, q2.NumericMetadataColumn)
@@ -61,6 +63,7 @@ def write_table(i, directory, table):
 def spiec_easi(
     table: biom.Table,
     metadata: q2.Metadata = None,
+    metadata_column: list = None,
     method: str = "glasso",
     lambda_min_ratio: float = 1e-3,
     nlambda: int = 20,
@@ -83,7 +86,8 @@ def spiec_easi(
             table_files.append(write_table(i, temp_dir, one_table))
             sample_ids = one_table.ids(axis='sample')
         if metadata:
-            metadata_tables = encode_metadata(sample_ids, metadata)
+            metadata_tables = encode_metadata(
+                sample_ids, metadata, metadata_column)
             metadata_columns = []
             for i, one_table in enumerate(metadata_tables, i + 1):
                 table_files.append(write_table(i, temp_dir, one_table))

diff --git a/q2_makarsa/plugin_setup.py b/q2_makarsa/plugin_setup.py
@@ -70,6 +70,7 @@ def _2(ff: NetworkFormat) -> Graph:
     inputs={"table": List[FeatureTable[Frequency]]},
     parameters={
         "metadata": Metadata,
+        "metadata_column": List[Str],
         "method": Str,
         "lambda_min_ratio": Float,
         "nlambda": Int,
@@ -94,6 +95,7 @@ def _2(ff: NetworkFormat) -> Graph:
     },
     parameter_descriptions={
         "metadata": "path to sample metadata",
+        "metadata_column": "only use this or these column or columns",
         "method": "Methods available for spieceasi,for example mb,glasso,slr",
         "lambda_min_ratio": (
             "Input parameter of spieceasi which represents "
@@ -130,6 +132,7 @@ def _2(ff: NetworkFormat) -> Graph:
     inputs={"table": FeatureTable[Frequency]},
     parameters={
         "metadata": Metadata,
+        "metadata_column": List[Str],
         "heterogeneous": Bool,
         "sensitive": Bool,
         "max_k": Int,
@@ -157,6 +160,7 @@ def _2(ff: NetworkFormat) -> Graph:
     },
     parameter_descriptions={
         "metadata": "path to sample metadata",
+        "metadata_column": "only use this or these column or columns",
         "heterogeneous": (
             "enable heterogeneous mode for multi-habitat or"
             "-protocol data with at least thousands of samples"