handle ancombc edge case with two classes + nan, also make filtered t…

…ables more manageable
clemente-lab · Aug 14, 2024 · e847c91 · e847c91
1 parent 1ee83ec
commit e847c91
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 4 deletions.
diff --git a/mmeds/snakemake/rules/common.smk b/mmeds/snakemake/rules/common.smk
@@ -37,7 +37,7 @@ def pairwise_splits(wildcards, tool, vars):
                     splits += expand("results/{var}/lefse_plot.{feature_table}.{var}.{subclass}.pdf",
                                     feature_table=tables, var=var, subclass=subclasses)
             elif tool == "ancombc":
-                splits += expand("differential_abundance/{var}/ancom-bc_barplot.{feature_table}.{var}.qzv",
+                splits += expand("differential_abundance/{var}/ancom-bc_barplot.{feature_table}-{var}-clean.{var}.qzv",
                                 feature_table=tables, var=var)
             continue
 

diff --git a/mmeds/snakemake/rules/differential_abundance.smk b/mmeds/snakemake/rules/differential_abundance.smk
@@ -1,6 +1,6 @@
 rule differential_abundance_ancom_bc:
     input:
-        feature_table = "tables/{table}.qza",
+        feature_table = "tables/{var}/{table}.qza",
         mapping_file = "tables/qiime_mapping_file.tsv"
     output:
         diffs = "differential_abundance/{var}/ancom-bc_diffs.{table}.{var}.qza",

diff --git a/mmeds/snakemake/rules/file_manipulation.smk b/mmeds/snakemake/rules/file_manipulation.smk
@@ -15,9 +15,26 @@ rule extract_feature_table_tsv:
         rm -f tables/{wildcards.table}.biom
         """
 
+rule extract_feature_table_tsv_class:
+    input:
+        "tables/{class}/{table}.qza"
+    output:
+        "tables/{class}/{table}.tsv"
+    conda:
+        "qiime2-2020.8.0"
+    shell:
+        """
+        unzip -jo {input} -d tables/tmp_unzip_{wildcards.table}
+        mv tables/tmp_unzip_{wildcards.table}/feature-table.biom tables/{wildcards.table}.biom
+        biom convert --to-tsv -i tables/{wildcards.table}.biom -o {output}
+        sed -i '1d;2s/^#//' {output}
+        rm -rf tables/tmp_unzip_{wildcards.table}
+        rm -f tables/{wildcards.table}.biom
+        """
+
 rule format_metadata_qiime_to_lefse:
     input:
-        feature_table = "tables/{table}.tsv",
+        feature_table = "tables/{class}/{table}.tsv",
         mapping_file = "tables/qiime_mapping_file.tsv"
     output:
         "tables/{class}/lefse_format.{table}.{class}.{subclass}.tsv"

diff --git a/mmeds/snakemake/rules/table_filtering.smk b/mmeds/snakemake/rules/table_filtering.smk
@@ -18,7 +18,7 @@ rule filter_table_to_two_classes:
         feature_table = "tables/{table}.qza",
         mapping_file = "tables/qiime_mapping_file.tsv"
     output:
-        "tables/{table}.{category}-{class1}-or-{class2}.qza"
+        "tables/{category}/{table}.{category}-{class1}-or-{class2}.qza"
     conda:
         "qiime2-2020.8.0"
     shell:
@@ -27,3 +27,18 @@ rule filter_table_to_two_classes:
         "--m-metadata-file {input.mapping_file} "
         "--p-where \"[{wildcards.category}]==\'{wildcards.class1}\' OR [{wildcards.category}]==\'{wildcards.class2}\'\" "
         "--o-filtered-table {output}"
+
+rule filter_nans:
+    input:
+        feature_table = "tables/{table}.qza",
+        mapping_file = "tables/qiime_mapping_file.tsv"
+    output:
+        "tables/{category}/{table}-{category}-clean.qza"
+    conda:
+        "qiime2-2020.8.0"
+    shell:
+        "qiime feature-table filter-samples "
+        "--i-table {input.feature_table} "
+        "--m-metadata-file {input.mapping_file} "
+        "--p-where \"[wildcards.category]!=\'nan\'\" "
+        "--o-filtered-table {output}"