Filtering taxa

dryrun_snakemake.bash

@@ -5,7 +5,7 @@ conda activate qiime2-2023.2
 
 set -xeuo pipefail
 
-if [[ -f config.yaml ]]; then
+if [[ ! -f config.yaml ]]; then
     echo "Must have a config.yaml to be able to run"
     exit 1
 fi

qiime2_config.yml

@@ -1,27 +1,27 @@
 all:
-    project_dir: "/mnt/isilon/microbiome/analysis/userid/16S_QIIME2/test"
+    project_dir: "/scr1/users/path/to/project"
     mux_dir: "multiplexed_fastq"
-    mapping: "test_mapping_file.tsv"
-    admin_email: "userid@email"
+    mapping: "mapping_table.tsv"
 
 #try revcomp if you are getting 0 reads when demultiplexing
 demux:
     mismatch: 0
-    revcomp: false
+    revcomp: true
 
 #~240 for V1V2, ~280 for V3V4
 denoise:
     trim_left_f: 0
     trunc_len_f: 240
     trim_left_r: 0
     trunc_len_r: 240
+    filter_taxa: "" # e.g. g__Cutibacterium, g__Anoxybacillus, f__Comamonadaceae, g__Acidovorax, g__Ralstonia
 #threads are now specificed in config.yaml
 
 taxonomy:
     classifier_fp: "/mnt/isilon/microbiome/analysis/biodata/silva/silva_138/silva-138-99-nb-classifier.qza"
 
 diversity:
-    sampling_depth: 100
+    sampling_depth: 1000
 #100 is just for testing, recommend change to 1000
 
 unassign:

rules/demux/dnabc.rules

@@ -33,7 +33,7 @@ rule run_dnabc:
             shell("""
             mkdir -p {params.demux_dir}
             touch {output.demux_fastq}
-            dnabc.py \
+            dnabc \
             --revcomp \
             --output-dir {params.demux_dir} \
             --mismatches {params.mismatch} \
@@ -46,7 +46,7 @@ rule run_dnabc:
             shell("""
             mkdir -p {params.demux_dir}
             touch {output.demux_fastq}
-            dnabc.py \
+            dnabc \
             --output-dir {params.demux_dir} \
             --mismatches {params.mismatch} \
             --manifest-file {output.manifest_fp} \

rules/denoise/denoise.rules

@@ -45,7 +45,7 @@ rule run_dada2:
         QIIME_OUTPUT_DIR + "/demux.qza"
     threads: 1 #should be overridden by config.yaml
     output:
-        feature_table = DENOISE_DIR + "/table_main.qza",
+        feature_table = DENOISE_DIR + "/table_raw.qza",
         rep_seq = DENOISE_DIR + "/representative-seqs.qza",
         stat = DENOISE_DIR + "/denoise_stats.qza"
     shell:
@@ -78,13 +78,35 @@ rule denoising_stats_export:
 
 rule filter_empty:
     input:
-        DENOISE_DIR + "/table_main.qza"
+        DENOISE_DIR + "/table_raw.qza"
     output:
-        DENOISE_DIR + "/table.qza"
+        DENOISE_DIR + "/table_filtered_empty.qza"
     shell:
         """
         qiime feature-table filter-samples \
         --i-table {input} \
         --p-min-frequency 1 \
         --o-filtered-table "{output}"
         """
+
+rule filter_taxa:
+    input:
+        feature_table = DENOISE_DIR + "/table_filtered_empty.qza",
+        classification_table = DENOISE_DIR + "/taxonomy/classification.qza"
+    params:
+        taxa_to_filter = config["denoise"]["filter_taxa"]
+    output:
+        DENOISE_DIR + "/table.qza"
+    run:
+        if params.taxa_to_filter:
+            shell("""
+            qiime taxa filter-table \
+            --i-table {input.feature_table} \
+            --i-taxonomy {input.classification_table} \
+            --p-exclude "{params.taxa_to_filter}" \
+            --o-filtered-table {output}
+            """)
+        else:
+            shell("""
+            cp {input.feature_table} {output}
+            """)

run_snakemake.bash

@@ -4,7 +4,7 @@
 #SBATCH -n 1
 #SBATCH --export=ALL
 #SBATCH --mail-user=$USER@chop.edu
-#SBATCH --mail-type=BEGIN,END,FAIL
+#SBATCH --mail-type=END,FAIL
 #SBATCH --no-requeue
 #SBATCH -t 72:00:00
 #SBATCH --output=slurm_%x_%j.out