add more temporary files to be removed when workflow finish

snakemodules/analysis.smk

@@ -44,6 +44,9 @@ if config["SICKLE"]=="no":
         group: "unmapped_analysis"
         notebook:
             os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
+
+
         
 
     if config["remove_contamination"]=="yes":
@@ -91,6 +94,7 @@ if config["SICKLE"]=="no":
             group: "unmapped_analysis_no_filtler_after_clean"
             notebook:
                 os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
             
 
 elif config["SICKLE"]=="yes":
@@ -139,6 +143,8 @@ elif config["SICKLE"]=="yes":
             group: "unmapped_analysis_before_filter"
             notebook:
                 os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
+
             
 
         rule email_unmapped_analysis_after_filter:
@@ -185,6 +191,8 @@ elif config["SICKLE"]=="yes":
             group: "unmapped_analysis_after_filter"
             notebook:
                 os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
+
             
 
     elif config["remove_contamination"]=="yes":
@@ -232,6 +240,7 @@ elif config["SICKLE"]=="yes":
             group: "unmapped_analysis_before_clean_filter"
             notebook:
                 os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
             
 
         rule email_unmapped_analysis_after_clean:
@@ -324,6 +333,7 @@ elif config["SICKLE"]=="yes":
             group: "unmapped_analysis_after_clean_filter"
             notebook:
                 os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/read_numbers_in_bar_graph.py.ipynb")
+
             
 
 
@@ -472,4 +482,3 @@ rule coverage_analysis:
     group: "coverage_analysis"
     notebook:
         os.path.join(WORKFLOW_PATH,"snakemodules/notebooks/contig_coverage_heatmap.py.ipynb")
-    

snakemodules/contigmap.smk

@@ -169,7 +169,7 @@ rule coordinate_sort_unsorted:
     input:
         WORK_DIR + "/{sample}/merged.bam" 
     output:
-        WORK_DIR + "/{sample}/non_ref_new.bam"
+        temp(WORK_DIR + "/{sample}/non_ref_new.bam")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
     resources:
@@ -192,7 +192,7 @@ rule index_sorted:
     input:
         WORK_DIR + "/{sample}/non_ref_new.bam"
     output:
-        WORK_DIR + "/{sample}/non_ref_new.bam.bai"
+        temp(WORK_DIR + "/{sample}/non_ref_new.bam.bai")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
     resources:

snakemodules/crop_remapped.smk

@@ -23,9 +23,9 @@ if config["SICKLE"]=="no":
             find_bam_file
             # INPUT_DIR + "/{sample}/{sample}.bam"     
         output:
-            WORK_DIR + "/{sample}/single.fastq", 
-            WORK_DIR + "/{sample}/paired.1.fastq", 
-            WORK_DIR + "/{sample}/paired.2.fastq", 
+            temp(WORK_DIR + "/{sample}/single.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.1.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.2.fastq"), 
             temp(WORK_DIR + "/{sample}/mates.bam"), 
             WORK_DIR + "/{sample}/POPINS_SAMPLE_INFO"
         params:
@@ -56,7 +56,7 @@ if config["SICKLE"]=="no":
             os.path.join(RESULTS_DIR, "read_numbers.png") if config["ANALYSIS"]=="yes" else [],
             mates = WORK_DIR + "/{sample}/mates.bam"
         output:
-            WORK_DIR + "/{sample}/non_ref.bam"
+            temp(WORK_DIR + "/{sample}/non_ref.bam")
         conda:
             os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
         resources:
@@ -78,9 +78,9 @@ elif config["SICKLE"]=="yes":
         input:
             INPUT_DIR + "/{sample}/{sample}.bam"     
         output:
-            WORK_DIR + "/{sample}/single.fastq", 
-            WORK_DIR + "/{sample}/paired.1.fastq", 
-            WORK_DIR + "/{sample}/paired.2.fastq", 
+            temp(WORK_DIR + "/{sample}/single.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.1.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.2.fastq"), 
             temp(WORK_DIR + "/{sample}/mates.bam"), 
             WORK_DIR + "/{sample}/POPINS_SAMPLE_INFO"
         params:
@@ -109,7 +109,7 @@ elif config["SICKLE"]=="yes":
             # os.path.join(RESULTS_DIR, "read_numbers.png") if config["ANALYSIS"]=="yes" else [],
             mates = WORK_DIR + "/{sample}/mates.bam"
         output:
-            WORK_DIR + "/{sample}/non_ref.bam"
+            temp(WORK_DIR + "/{sample}/non_ref.bam")
         conda:
             os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
         resources:

snakemodules/crop_unmapped.smk

@@ -21,9 +21,9 @@ if config["SICKLE"]=="no":
             find_bam_file
             # INPUT_DIR + "/{sample}/{sample}.bam"       
         output:
-            WORK_DIR + "/{sample}/single.fastq", 
-            WORK_DIR + "/{sample}/paired.1.fastq", 
-            WORK_DIR + "/{sample}/paired.2.fastq", 
+            temp(WORK_DIR + "/{sample}/single.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.1.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.2.fastq"), 
             temp(WORK_DIR + "/{sample}/mates.bam"), 
             WORK_DIR + "/{sample}/POPINS_SAMPLE_INFO"
         params:
@@ -54,7 +54,7 @@ if config["SICKLE"]=="no":
             os.path.join(RESULTS_DIR, "read_numbers.png") if config["ANALYSIS"]=="yes" else [],
             mates = WORK_DIR + "/{sample}/mates.bam"
         output:
-            WORK_DIR + "/{sample}/non_ref.bam"
+            temp(WORK_DIR + "/{sample}/non_ref.bam")
         conda:
             os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
         resources:
@@ -76,9 +76,9 @@ elif config["SICKLE"]=="yes":
         input:
             INPUT_DIR + "/{sample}/{sample}.bam"       
         output:
-            WORK_DIR + "/{sample}/single.fastq", 
-            WORK_DIR + "/{sample}/paired.1.fastq", 
-            WORK_DIR + "/{sample}/paired.2.fastq", 
+            temp(WORK_DIR + "/{sample}/single.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.1.fastq"), 
+            temp(WORK_DIR + "/{sample}/paired.2.fastq"), 
             temp(WORK_DIR + "/{sample}/mates.bam"), 
             WORK_DIR + "/{sample}/POPINS_SAMPLE_INFO"
         params:
@@ -107,7 +107,7 @@ elif config["SICKLE"]=="yes":
             # os.path.join(RESULTS_DIR, "read_numbers.png") if config["ANALYSIS"]=="yes" else [],
             mates = WORK_DIR + "/{sample}/mates.bam"
         output:
-            WORK_DIR + "/{sample}/non_ref.bam"
+            temp(WORK_DIR + "/{sample}/non_ref.bam")
         resources:
             mem_per_thread = resources["samtools_multithread"]["mem_per_thread"],
             mem_mb = lambda wildcards, input, threads, attempt: resources["samtools_multithread"]["mem_per_thread"] * threads,

snakemodules/genotype.smk

@@ -28,7 +28,7 @@ rule popins2_genotype:
     input:
         rules.sort_vcf.output
     output:
-        WORK_DIR + "/{sample}/insertions.vcf"
+        temp(WORK_DIR + "/{sample}/insertions.vcf")
     resources:
         mem_mb = resources["standard_2G"]["mem"],
         runtime = resources["standard_2G"]["time"]

snakemodules/kraken.smk

@@ -8,15 +8,15 @@ rule kraken_map:
         p2 = WORK_DIR + "/{sample}/paired.2.fastq"
     output:
         single_report = WORK_DIR + "/{sample}/contaminate_info/single_report.txt",
-        single_output = WORK_DIR + "/{sample}/contaminate_info/single_output.txt",
-        u_single = WORK_DIR + "/{sample}/contaminate_info/useq_single.fq",
-        c_single = WORK_DIR + "/{sample}/contaminate_info/cseq_single.fq",
+        single_output = temp(WORK_DIR + "/{sample}/contaminate_info/single_output.txt"),
+        u_single = temp(WORK_DIR + "/{sample}/contaminate_info/useq_single.fq"),
+        c_single = temp(WORK_DIR + "/{sample}/contaminate_info/cseq_single.fq"),
         paired_report = WORK_DIR + "/{sample}/contaminate_info/paired_report.txt",
-        paired_output = WORK_DIR + "/{sample}/contaminate_info/paired_output.txt",
-        c_paired_1 = WORK_DIR + "/{sample}/contaminate_info/cseqs_1.fq",
-        c_paired_2 = WORK_DIR + "/{sample}/contaminate_info/cseqs_2.fq",
-        u_paired_1 = WORK_DIR + "/{sample}/contaminate_info/useqs_1.fq",
-        u_paired_2 = WORK_DIR + "/{sample}/contaminate_info/useqs_2.fq"
+        paired_output = temp(WORK_DIR + "/{sample}/contaminate_info/paired_output.txt"),
+        c_paired_1 = temp(WORK_DIR + "/{sample}/contaminate_info/cseqs_1.fq"),
+        c_paired_2 = temp(WORK_DIR + "/{sample}/contaminate_info/cseqs_2.fq"),
+        u_paired_1 = temp(WORK_DIR + "/{sample}/contaminate_info/useqs_1.fq"),
+        u_paired_2 = temp(WORK_DIR + "/{sample}/contaminate_info/useqs_2.fq")
     singularity:
         "docker://staphb/kraken2:latest"
     conda:
@@ -63,9 +63,9 @@ rule obtain_classified_human_reads:
         paired_fq2 = rules.kraken_map.output.c_paired_2,
         script= os.path.join(WORKFLOW_PATH,config["python_script"])
     output:
-        hc_single = WORK_DIR + "/{sample}/contaminate_info/human_classified_single.fastq",
-        hc_paired_1 = WORK_DIR + "/{sample}/contaminate_info/human_classified_paired_1.fastq",
-        hc_paired_2 = WORK_DIR + "/{sample}/contaminate_info/human_classified_paired_2.fastq"
+        hc_single = temp(WORK_DIR + "/{sample}/contaminate_info/human_classified_single.fastq"),
+        hc_paired_1 = temp(WORK_DIR + "/{sample}/contaminate_info/human_classified_paired_1.fastq"),
+        hc_paired_2 = temp(WORK_DIR + "/{sample}/contaminate_info/human_classified_paired_2.fastq")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/biopython.yml")
     resources:
@@ -146,7 +146,7 @@ rule samtools_remap_classified_human:
         sam = rules.bwa_remap_classified_human.output.remapped_sam
     output:
         remapped_unsorted = temp(WORK_DIR + "/{sample}/contaminate_info/remapped_human_unsorted.bam"),
-        remapped_bam = WORK_DIR + "/{sample}/contaminate_info/remapped.bam"
+        remapped_bam = temp(WORK_DIR + "/{sample}/contaminate_info/remapped.bam")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
     resources:
@@ -169,7 +169,7 @@ rule index_reads:
     input:
         bam = rules.samtools_remap_classified_human.output.remapped_bam
     output:
-        bai = WORK_DIR + "/{sample}/contaminate_info/remapped.bam.bai"
+        bai = temp(WORK_DIR + "/{sample}/contaminate_info/remapped.bam.bai")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
     resources:
@@ -191,9 +191,9 @@ rule crop_remapped:
         bam = rules.samtools_remap_classified_human.output.remapped_bam,
         bai = rules.index_reads.output.bai
     output:
-        single = WORK_DIR + "/{sample}/remapped_single.fastq", 
-        p1 = WORK_DIR + "/{sample}/remapped_paired.1.fastq", 
-        p2 = WORK_DIR + "/{sample}/remapped_paired.2.fastq", 
+        single = temp(WORK_DIR + "/{sample}/remapped_single.fastq"), 
+        p1 = temp(WORK_DIR + "/{sample}/remapped_paired.1.fastq"), 
+        p2 = temp(WORK_DIR + "/{sample}/remapped_paired.2.fastq"), 
         unsorted = temp(WORK_DIR + "/{sample}/remapped_mates.unsorted.bam")
     resources:
         mem_mb = resources["standard_2G"]["mem"],
@@ -220,7 +220,7 @@ rule remapping_samsort_mates:
     input:
         WORK_DIR + "/{sample}/remapped_mates.unsorted.bam"
     output:
-        WORK_DIR + "/{sample}/remapped_mates.bam"
+        temp(WORK_DIR + "/{sample}/remapped_mates.bam")
     conda:
         os.path.join(WORKFLOW_PATH,"snakemodules/envs/samtools.yml")
     resources:
@@ -242,7 +242,7 @@ rule merge_set_mate:
         non_ref = WORK_DIR + "/{sample}/non_ref.bam",
         rm_ref = WORK_DIR + "/{sample}/remapped_mates.bam"
     output:
-        WORK_DIR + "/{sample}/remapped_non_ref.bam"
+        temp(WORK_DIR + "/{sample}/remapped_non_ref.bam")
     resources:
         mem_mb = resources["standard_2G"]["mem"],
         runtime = resources["standard_2G"]["time"]
@@ -270,9 +270,9 @@ rule contaminate_removed:
         u_p1 = rules.kraken_map.output.u_paired_1,
         u_p2 = rules.kraken_map.output.u_paired_2
     output:
-        single_clean = WORK_DIR + "/{sample}/single_clean.fastq",
-        p1_clean = WORK_DIR + "/{sample}/paired_1_clean.fastq",
-        p2_clean = WORK_DIR + "/{sample}/paired_2_clean.fastq"
+        single_clean = temp(WORK_DIR + "/{sample}/single_clean.fastq"),
+        p1_clean = temp(WORK_DIR + "/{sample}/paired_1_clean.fastq"),
+        p2_clean = temp(WORK_DIR + "/{sample}/paired_2_clean.fastq")
     resources:
         mem_mb = resources["standard_2G"]["mem"],
         runtime = resources["standard_2G"]["time"]

snakemodules/minia.smk

@@ -8,7 +8,7 @@ if config["SICKLE"]=="yes":
                 pair_1 = WORK_DIR + "/{sample}/sickle.paired_1_clean.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/sickle.paired_2_clean.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             conda:
                 os.path.join(WORKFLOW_PATH,"snakemodules/envs/py27.yml")
             resources:
@@ -36,7 +36,7 @@ if config["SICKLE"]=="yes":
                 pair_1 = WORK_DIR + "/{sample}/sickle.paired.1.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/sickle.paired.2.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             conda:
                 os.path.join(WORKFLOW_PATH,"snakemodules/envs/py27.yml")
             resources:
@@ -64,7 +64,7 @@ elif config["SICKLE"]=="no":
                 pair_1 = WORK_DIR + "/{sample}/paired_1_clean.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/paired_2_clean.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             conda:
                 os.path.join(WORKFLOW_PATH,"snakemodules/envs/py27.yml")
             resources:
@@ -92,7 +92,7 @@ elif config["SICKLE"]=="no":
                 pair_1 = WORK_DIR + "/{sample}/paired.1.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/paired.2.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             conda:
                 os.path.join(WORKFLOW_PATH,"snakemodules/envs/py27.yml")
             resources:

snakemodules/position.smk

@@ -8,7 +8,7 @@ if config["remove_contamination"] == 'yes':
             os.path.join(RESULTS_DIR, "heatmap_coverage.png") if config["ANALYSIS"]=="yes" else [],
             WORK_DIR + "/{sample}/remapped_non_ref.bam"
         output:
-            WORK_DIR + "/{sample}/locations.txt"
+            temp(WORK_DIR + "/{sample}/locations.txt")
         resources:
             mem_mb = resources["standard_4G"]["mem"],
             runtime = resources["standard_4G"]["time"]
@@ -34,7 +34,7 @@ elif config["remove_contamination"] == 'no':
             os.path.join(RESULTS_DIR, "heatmap_coverage.png") if config["ANALYSIS"]=="yes" else [],
             expand(WORK_DIR + "/{s}/{f}", s=SAMPLES, f=["non_ref_new.bam","non_ref.bam"])
         output:
-            WORK_DIR + "/{sample}/locations.txt"
+            temp(WORK_DIR + "/{sample}/locations.txt")
         resources:
             mem_mb = resources["standard_4G"]["mem"],
             runtime = resources["standard_4G"]["time"]

snakemodules/velvet.smk

@@ -7,7 +7,7 @@ if config["SICKLE"]=="yes":
                 pair_1 = WORK_DIR + "/{sample}/sickle.paired_1_clean.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/sickle.paired_2_clean.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             params:
                 kmerlength = config["kmerlength"]
             conda:
@@ -37,7 +37,7 @@ if config["SICKLE"]=="yes":
                 pair_1 = WORK_DIR + "/{sample}/sickle.paired.1.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/sickle.paired.2.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             params:
                 kmerlength = config["kmerlength"]
             conda:
@@ -67,7 +67,7 @@ elif config["SICKLE"]=="no":
                 pair_1 = WORK_DIR + "/{sample}/paired_1_clean.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/paired_2_clean.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             params:
                 kmerlength = config["kmerlength"]
             conda:
@@ -97,7 +97,7 @@ elif config["SICKLE"]=="no":
                 pair_1 = WORK_DIR + "/{sample}/paired.1.fastq", 
                 pair_2 = WORK_DIR + "/{sample}/paired.2.fastq"
             output:
-                WORK_DIR + "/{sample}/{assemblr}.contigs.fa"
+                temp(WORK_DIR + "/{sample}/{assemblr}.contigs.fa")
             params:
                 kmerlength = config["kmerlength"]
             conda: