Merge pull request PennChopMicrobiomeProgram#1 from ctanes/master

Cleaned diversity rules and took out report
ctanes · Dec 9, 2021 · 0a4f21b · 0a4f21b
2 parents 9e86dc8 + d62cdd1
commit 0a4f21b
Show file tree

Hide file tree

Showing 11 changed files with 167 additions and 1,140 deletions.
diff --git a/README.md b/README.md
@@ -113,10 +113,4 @@ snakemake \
 - Representative sequences (fasta)
 #### Output
 - Vsearch report (tsv) customized to be like BLAST results (see config.yml)
-- Vsearch list of representative sequences that aligned (fasta)
-
-### Basic Bioinformatics Report
-#### Input
-- QIIME2 compatible mapping file and output from diversity calculation 
-#### Output
-- Basic Bioinformatics Report containging heatmap, relative proportion bar graph, alpha diversity plots, beta diversity plots, and per sample read counts in HTML format.
+- Vsearch list of representative sequences that aligned (fasta)
diff --git a/Snakefile b/Snakefile
@@ -13,12 +13,12 @@ trunc_len_f = config["denoise"]["trunc_len_f"]
 trim_left_r = config["denoise"]["trim_left_r"]
 trunc_len_r = config["denoise"]["trunc_len_r"]
 
-DENOISE_DIR = (QIIME_OUTPUT_DIR + "/denoise" +
-                "_fwd_" + str(trim_left_f) + "-" + str(trunc_len_f) +
-                "_rev_" + str(trim_left_r) + "-" + str(trunc_len_r)
-                    )
+DENOISE_DIR = (QIIME_OUTPUT_DIR + "/denoise-results")
+
 CORE_METRIC_DIR = (DENOISE_DIR + "/core-metrics" +
-                    "_sampling_depth_" + str(config["diversity"]["sampling_depth"]))
+		  "_sampling_depth_" + str(config["diversity"]["sampling_depth"]))
+
+CORE_METRIC_UNRAREFIED_DIR = (DENOISE_DIR + "/core-metrics-unrarefied")
 
 include: "rules/targets/targets.rules"
 include: "rules/demux/dnabc.rules"
@@ -29,7 +29,6 @@ include: "rules/taxonomy/taxonomy.rules"
 include: "rules/tree/tree.rules"
 include: "rules/diversity/diversity.rules"
 include: "rules/unassign/unassign.rules"
-include: "rules/report/report.rules"
 include: "rules/dada2_species/dada2.rules"
 include: "rules/vsearch/vsearch.rules"
 

diff --git a/config.yml b/config.yml
@@ -20,19 +20,11 @@ taxonomy:
     classifier_fp: "/home/danielsg/16S_QIIME2/QIIME2_data/gg-13-8-99-nb-classifier.qza"
 
 diversity:
-    sampling_depth: 30
+    sampling_depth: 1000
 
 unassign:
     unassigner_species_fp: "/home/danielsg/16S_QIIME2/unassign_data/unassigner_species.fasta"
 
-report:
-    study_group_var: "investigator"
-    min_reads: 5
-    richness_subsample_size: 20
-    rmd: "/home/danielsg/16S_QIIME2/scripts/Basic_Bioinformatics_Report.Rmd"
-    R_helper: "/home/danielsg/16S_QIIME2/scripts/helper_functions.R"
-    report_rdata: "/home/danielsg/16S_QIIME2/test/report.RData" # rmd can only have one output so have to put this here
-
 dada2:
     rscript: "/home/danielsg/16S_QIIME2/scripts/dada2.R"
     species_training_set: "/home/danielsg/16S_QIIME2/dada2_data/rdp_species_assignment_16.fa.gz"

diff --git a/dryrun_snakemake.bash b/dryrun_snakemake.bash
@@ -7,8 +7,8 @@
 #$ -j y
 
 #Uncomment the next two lines if you want to 'qsub' this script
-source ~/.bashrc #needed to make "conda" command to work
-conda activate qiime2-snakemake
+#source ~/.bashrc #needed to make "conda" command to work
+#conda activate qiime2-snakemake
 
 set -xeuo pipefail
 

diff --git a/rules/denoise/denoise.rules b/rules/denoise/denoise.rules
@@ -46,7 +46,7 @@ rule run_dada2:
     params:
         n_threads = config["denoise"]["n_threads"]
     output:
-        feature_table = DENOISE_DIR + "/table.qza",
+        feature_table = DENOISE_DIR + "/table_main.qza",
         rep_seq = DENOISE_DIR + "/representative-seqs.qza",
         stat = DENOISE_DIR + "/denoise_stats.qza"
     shell:
@@ -62,3 +62,16 @@ rule run_dada2:
         --o-representative-sequences {output.rep_seq} \
         --o-denoising-stats {output.stat}
         """ 
+
+rule filter_empty:
+    input:
+        DENOISE_DIR + "/table_main.qza"
+    output:
+        DENOISE_DIR + "/table.qza"
+    shell:
+        """
+        qiime feature-table filter-samples \
+        --i-table {input} \
+        --p-min-frequency 1 \
+        --o-filtered-table "{output}"
+        """
diff --git a/rules/diversity/diversity.rules b/rules/diversity/diversity.rules
@@ -134,3 +134,129 @@ rule get_diversity:
         --o-bray-curtis-emperor {output.bc_emperor} 
         """
 
+rule get_uu_unrarefied:
+     input:
+         rooted_tree = DENOISE_DIR + "/rooted-tree.qza",
+         table = DENOISE_DIR + "/table.qza"
+     output:
+         CORE_METRIC_UNRAREFIED_DIR + "/unweighted_unifrac_distance_matrix.qza"
+     shell:
+         """
+         qiime diversity beta-phylogenetic \
+      	 --i-phylogeny {input.rooted_tree} \
+      	 --i-table {input.table} \
+      	 --p-metric unweighted_unifrac \
+      	 --o-distance-matrix {output}
+         """
+
+rule move_uu_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/uu/distance-matrix.tsv"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/uu"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/uu_unrarefied.tsv"
+    shell:
+        """
+        mv {input} {output}
+        rmdir {params}
+        """
+
+rule uu_convert_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/unweighted_unifrac_distance_matrix.qza"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/uu"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/uu/distance-matrix.tsv"
+    shell:
+        """
+        qiime tools export \
+        --input-path {input} \
+        --output-path {params}
+        """
+
+rule get_wu_unrarefied:
+     input:
+         rooted_tree = DENOISE_DIR + "/rooted-tree.qza",
+         table = DENOISE_DIR + "/table.qza"
+     output:
+         CORE_METRIC_UNRAREFIED_DIR + "/weighted_unifrac_distance_matrix.qza"
+     shell:
+         """
+         qiime diversity beta-phylogenetic \
+         --i-phylogeny  {input.rooted_tree} \
+      	 --i-table {input.table} \
+      	 --p-metric weighted_unifrac \
+      	 --o-distance-matrix {output}
+         """
+
+rule move_wu_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/wu/distance-matrix.tsv"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/wu"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/wu_unrarefied.tsv"
+    shell:
+        """
+        mv {input} {output}
+        rmdir {params}
+        """
+
+rule wu_convert_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/weighted_unifrac_distance_matrix.qza"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/wu"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/wu/distance-matrix.tsv"
+    shell:
+        """
+        qiime tools export \
+        --input-path {input} \
+        --output-path {params}
+        """
+
+
+rule get_faith_unrarefied:
+     input:
+         rooted_tree = DENOISE_DIR + "/rooted-tree.qza",
+         table = DENOISE_DIR + "/table.qza"
+     output:
+         CORE_METRIC_UNRAREFIED_DIR + "/faith_pd_vector.qza"
+     shell:
+         """
+         qiime diversity alpha-phylogenetic \
+      	 --i-phylogeny  {input.rooted_tree} \
+      	 --i-table {input.table} \
+      	 --p-metric faith_pd \
+      	 --o-alpha-diversity {output}
+         """
+
+rule move_faith_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith/alpha-diversity.tsv"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith_pd_unrarefied.tsv"
+    shell:
+        """
+        mv {input} {output}
+        rmdir {params}
+        """
+
+rule faith_convert_unrarefied:
+    input:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith_pd_vector.qza"
+    params:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith"
+    output:
+        CORE_METRIC_UNRAREFIED_DIR + "/faith/alpha-diversity.tsv"
+    shell:
+        """
+        qiime tools export \
+        --input-path {input} \
+        --output-path {params}
+        """
diff --git a/rules/report/report.rules b/rules/report/report.rules
diff --git a/rules/targets/targets.rules b/rules/targets/targets.rules
@@ -14,12 +14,13 @@ TARGET_TREE = [str(DENOISE_DIR + "/rooted-tree.qza")]
 
 TARGET_DIVERSITY = [str(CORE_METRIC_DIR + "/faith_pd.tsv"),
                     str(CORE_METRIC_DIR + "/uu.tsv"),
-                    str(CORE_METRIC_DIR + "/wu.tsv")]
+                    str(CORE_METRIC_DIR + "/wu.tsv"),
+		    str(CORE_METRIC_UNRAREFIED_DIR + "/faith_pd_unrarefied.tsv"),
+		    str(CORE_METRIC_UNRAREFIED_DIR + "/uu_unrarefied.tsv"),
+		    str(CORE_METRIC_UNRAREFIED_DIR + "/wu_unrarefied.tsv")]
 
 TARGET_UNASSIGN = [str(DENOISE_DIR + "/representative_seq_fasta/unassigned/unassigner_output.tsv")]
 
-TARGET_REPORT = [str(CORE_METRIC_DIR + "/basic_bioinformatics_report/Basic_Bioinformatics_Report.html")]
-
 TARGET_DADA2SP = [str(DENOISE_DIR + "/representative_seq_fasta/dada2/dada2_species_assignments.tsv"),
                   str(DENOISE_DIR + "/representative_seq_fasta/dada2/dada2.RData")]
 
@@ -37,8 +38,7 @@ TARGET_ALL = (
     TARGET_TAXONOMY +
     TARGET_TREE +
     TARGET_DIVERSITY +
-    TARGET_UNASSIGN +
-    TARGET_REPORT +
-    TARGET_DADA2SP +
-    TARGET_VSEARCH
+    TARGET_UNASSIGN #+
+    #TARGET_DADA2SP +
+    #TARGET_VSEARCH
 )
diff --git a/run_snakemake.bash b/run_snakemake.bash
@@ -1,14 +1,17 @@
 #!/usr/bin/env bash
 
-#$ -cwd
-#$ -r n
-#$ -V
-#$ -l h_vmem=2G
-#$ -j y
+#SBATCH --mem=2G
+#SBATCH -n 1
+#SBATCH --export=ALL
+#SBATCH [email protected]
+#SBATCH --mail-type=END,FAIL
+#SBATCH --no-requeue
+#SBATCH -t 12:00:00
+#SBATCH --output=slurm_%x_%j.out
 
 #Uncomment the next two lines if you want to 'qsub' this script
-source ~/.bashrc #needed to make "conda" command to work
-conda activate qiime2-snakemake
+#source ~/.bashrc #needed to make "conda" command to work
+#conda activate qiime2-snakemake
 
 set -xeuo pipefail
 
@@ -28,4 +31,4 @@ snakemake \
     --notemp \
     --printshellcmds \
     --cluster \
-    "qsub -cwd -r n -V -l h_vmem={cluster.h_vmem} -l mem_free={cluster.mem_free} -pe smp {threads}"
+    "sbatch --mem=10G -t 12:00:00 -n 4 --export=ALL --no-requeue"