Enhancement kb

mmeds/config.py

@@ -291,6 +291,12 @@
             "plot_max_rows",
             "match_string"
         ]
+    },
+    "picrust2": {
+        "parameters": [
+            "tables"  # this is going to always be 'asv_table.qza' and 'rep_seqs_table.qza'  TODO: default parameters?
+        ],
+        "optional_parameters": []
     }
 }
 

mmeds/snakemake/rules/file_manipulation.smk

@@ -12,6 +12,41 @@ rule extract_feature_table_tsv:
         "extract_feature_table.sh "
         "{input} "
         "{output} "
+        "tsv "
+        "tables/tmp_unzip_{wildcards.table}"
+
+rule extract_feature_table_biom:
+    """ Remove feature table biom file from qza archive """
+    input: 
+        "tables/{table}.qza"
+    output:
+        "tables/{table}.biom"
+    wildcard_constraints:
+        table = "[^/]+"
+    conda:
+        "mmeds_test"
+    shell:
+        "extract_feature_table.sh "
+        "{input} "
+        "{output} "
+        "biom "
+        "tables/tmp_unzip_{wildcards.table}"
+
+rule extract_feature_table_fasta:
+    """ Remove underlying fasta data from qza archive """
+    input: 
+        "tables/{table}.qza"
+    output:
+        "tables/{table}.fasta"
+    wildcard_constraints:
+        table = "[^/]+"
+    conda:
+        "mmeds_test"
+    shell:
+        "extract_feature_table.sh "
+        "{input} "
+        "{output} "
+        "fasta "
         "tables/tmp_unzip_{wildcards.table}"
 
 rule extract_feature_table_tsv_class:

mmeds/snakemake/rules/functional_annotation.smk

@@ -0,0 +1,15 @@
+rule picrust2_core:
+    """ Generate picrust2 tables for input feature table """
+    input:
+        rep_seqs = "tables/rep_seqs_table.fasta",
+        biom_feature_table = "tables/asv_table.biom"
+    output:
+        directory("picrust2_out")
+    conda:
+        "picrust2"
+    shell:
+        "picrust2_pipeline.py "
+        "-s {input.rep_seqs} "
+        "-i {input.biom_feature_table} "
+        "-o {output} "
+        "--stratified"

mmeds/snakemake/workflows/picrust2.Snakefile

@@ -0,0 +1,11 @@
+configfile: "config_file.yaml"
+report: "report.rst"
+
+include: "{snakemake_dir}/common.smk"
+include: "{snakemake_dir}/file_manipulation.smk"
+include: "{snakemake_dir}/table_filtering.smk"
+include: "{snakemake_dir}/functional_annotation.smk"
+
+rule results:
+    input:
+        directory("picrust2_out")

mmeds/tests/unit/test_snakemake.py

@@ -6,6 +6,7 @@
 from mmeds.util import setup_environment
 from mmeds.logging import Logger
 
+
 class SnakemakeTests(TestCase):
     """ Test running snakemake analyses """
     @classmethod
@@ -36,13 +37,11 @@ def run_snakemake(self, path):
         run(f"rm -f {dag}; rm -f {rulegraph}; rm -f {snakefile}", shell=True)
         return 0
 
-
     def test_a_core_pipeline_taxonomic(self):
         """ Test snakemake standard pipeline analysis """
         result = self.run_snakemake(Path(fig.TEST_SNAKEMAKE_DIR) / "core_pipeline_taxonomic")
         self.assertEquals(result, 0)
 
-
     def test_b_lefse(self):
         """ Test snakemake lefse analysis """
         result = self.run_snakemake(Path(fig.TEST_SNAKEMAKE_DIR) / "lefse")
@@ -53,3 +52,8 @@ def test_c_lefse_failure(self):
         """ Test analysis with undefined snakemake method """
         result = self.run_snakemake(Path(fig.TEST_SNAKEMAKE_DIR) / "lefse_failure")
         self.assertEquals(result, 0)
+
+    def test_d_picrust2(self):
+        """ Test snakemake picrust2 analysis """
+        result = self.run_snakemake(Path(fig.TEST_SNAKEMAKE_DIR) / "picrust2")
+        self.assertEquals(result, 0)

scripts/extract_feature_table.sh

@@ -4,8 +4,21 @@
 
 in_file=$1
 out_file=$2
-tmp_dir_name="${3:-tmp_unzip}"
+file_type="${3:-tsv}" # tsv, biom, or fasta
+tmp_dir_name="${4:-tmp_unzip}"
 unzip -joq $in_file -d $tmp_dir_name
-biom convert --to-tsv -i $tmp_dir_name/feature-table.biom -o $out_file
-sed -i '1d;2s/^#//' $out_file
-rm -rf $tmp_dir_name
+
+if [[ "$file_type" == "tsv" ]]; then
+    biom convert --to-tsv -i $tmp_dir_name/feature-table.biom -o $out_file
+    sed -i '1d;2s/^#//' $out_file
+    rm -rf $tmp_dir_name
+elif [[ "$file_type" == "biom" ]]; then
+    cp $tmp_dir_name/feature-table.biom $out_file
+    rm -rf $tmp_dir_name
+elif [[ "$file_type" == "fasta" ]]; then
+    cp $tmp_dir_name/dna-sequences.fasta $out_file
+    rm -rf $tmp_dir_name
+else
+    echo "Invalid file_type argument"
+fi
+

setup.py

@@ -75,6 +75,7 @@ def run(self):
           'pygments==2.8.1',
           'pymysql==1.0.2',
           'pytest-cov==2.11.1',
+          'pytest==8.3.5',
           'pytest-pudb==0.7.0',
           'pytest==8.3.5',
           'pytidylib==0.3.2',

test_files/test_snakemake/picrust2/picrust2.Snakefile

@@ -0,0 +1,11 @@
+configfile: "test_picrust2_config.yaml"
+report: "report.rst"
+
+include: "snakemake_dir/common.smk"
+include: "snakemake_dir/file_manipulation.smk"
+include: "snakemake_dir/table_filtering.smk"
+include: "snakemake_dir/functional_annotation.smk"
+
+rule results:
+    input:
+        directory("picrust2_out")

test_files/test_snakemake/picrust2/test_picrust2_config.yaml

@@ -0,0 +1,4 @@
+tables:
+  - asv_table
+  - rep_seqs_table
+