Code Review + code and logs harmonization

frangiPANe/snakefiles/abyss.smk

@@ -15,13 +15,13 @@ rule abyss:
         f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - bam file : {{input.bam_file}}
+|   - bam file : {input.bam_file}
 |-- Output:
-|   - Abyss dir: {f"{frangipane_obj.path_contigs}"}
+|   - Abyss dir: {frangipane_obj.path_contigs}
 |-- Others infos :
-|   - Threads : {{threads}}
-|   - LOG error: {{log.error}}
-|   - other_options : {{params.other_options}}
+|   - Threads : {threads}
+|   - LOG output : {log.output} 
+|   - other_options : {params.other_options}
 --------------------
 """
     singularity:

frangiPANe/snakefiles/abyss_merge.smk

@@ -5,9 +5,7 @@ rule abyss_merge:
         merged_fasta = f"{frangipane_obj.path_merged_contigs}/all-contigs.fa"
     log:
         output =  f"{frangipane_obj.path_logs}/post-abyss/abyss-merged.log.o",
-        error =  f"{frangipane_obj.path_logs}/post-abyss/abyss-merged.log.e",
-    params:
-        dir_merged = f"{frangipane_obj.path_merged_contigs}"
+        error =  f"{frangipane_obj.path_logs}/post-abyss/abyss-merged.log.e"
     message:
         f"""--------------------
 -Running {{rule}}
@@ -15,6 +13,8 @@ rule abyss_merge:
 |   - dir : f"{frangipane_obj.path_merged_contigs}"
 |-- Output:
 |   - fasta merged file : {{output.merged_fasta}}
+|-- Other infos:
+|   - LOG output : {log.output} 
 --------------------
 """
     run:

frangiPANe/snakefiles/bwa_index.smk

@@ -16,7 +16,6 @@ rule bwa_index:
 |-- Output:
 |   - index_file: {{output}} 
 |-- Others infos :
-|   -LOG error: {{log.error}}
 |   -LOG output: {{log.output}}
 |   -Other options: {{params.other_options}}
 --------------------

frangiPANe/snakefiles/bwa_mem.smk

@@ -26,7 +26,6 @@ rule bwa_mem:
 |   - bam file: {{output}}
 |-- Others infos :
 |   - Threads : {{threads}}
-|   - LOG error: {{log.error}}
 |   - LOG output: {{log.output}}
 |   - Other options bwa: {{params.other_options_bwaMem}}
 |   - Other options samtools sort: {{params.other_options_samtoolsSort}}

frangiPANe/snakefiles/cdhit.smk

@@ -17,8 +17,6 @@ rule cdhit:
 |-- Output:
 |   - Fasta file: {{output.fasta_file}} 
 |-- Others infos :
-|   -Threads : 6
-|   -LOG error: {{log.error}}
 |   -LOG output: {{log.output}}
 |   -Other options: {{params.other_options}}
 --------------------

frangiPANe/snakefiles/create_panref.smk

@@ -16,7 +16,6 @@ rule create_panref:
 |-- Output:
 |   - Fasta file: {{output.fasta_file}} 
 |-- Others infos :
-|   -LOG error: {{log.error}}
 |   -LOG output: {{log.output}}
 --------------------
 """

frangiPANe/snakefiles/fasta_stats.smk

@@ -4,8 +4,8 @@ rule fasta_stats:
     output: 
         stat_file = f"{frangipane_obj.path_stat_sum}/03_all_fasta-filtered-stats.csv"
     log:
-        output =  f"{frangipane_obj.path_logs}/report/fasta_stats.log.o",
-        error =  f"{frangipane_obj.path_logs}/report/fasta_stats.log.e"
+        output =  f"{frangipane_obj.path_logs}/stats/fasta_stats.o",
+        error =  f"{frangipane_obj.path_logs}/stats/fasta_stats.e"
     params:
         dir = f"{frangipane_obj.path_filtered_contigs}",
         other_option = frangipane_obj.get_config_value("PARAMS","assembly_stats")
@@ -13,9 +13,9 @@ rule fasta_stats:
          f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - Fasta dir : f"{{params.dir}}"
+|   - Fasta dir : {{params.dir}}
 |-- Output:
-|   - stat file: f"{{output.stat_file}}"
+|   - stat file: {{output.stat_file}}
 |-- Others infos :
 |   -LOG error: {{log.error}}
 --------------------
@@ -27,6 +27,6 @@ rule fasta_stats:
     shell:
         """
         cmd="assembly-stats {params.other_option} {params.dir}/*.fa";
-        (echo $cmd) > {log.error};
+        (echo $cmd) > {log.output};
         ($cmd) 1>> {output.stat_file} 2>> {log.error};
         """   
\ No newline at end of file

frangiPANe/snakefiles/fasta_stats2.smk

@@ -4,17 +4,17 @@ rule fasta_stats2:
     output: 
         stat_file = f"{frangipane_obj.path_stat_sum}/04_all_contigs-merged-stats.csv"
     log:
-        output =  f"{ frangipane_obj.path_logs}/fasta_stats/fasta_stats_contigs_merged.log.o",
-        error =  f"{ frangipane_obj.path_logs}/fasta_stats/fasta_stats_contigs_merged.log.e"
+        output =  f"{ frangipane_obj.path_logs}/stats/fasta_stats_contigs_merged.o",
+        error =  f"{ frangipane_obj.path_logs}/stats/fasta_stats_contigs_merged.e"
     params:
-        other_option = frangipane_obj.get_config_value("PARAMS","assembly_stats2") #config['PARAMS']['assembly_stats2'] if "assembly_stats2" in config['PARAMS'] else ""
+        other_option = frangipane_obj.get_config_value("PARAMS","assembly_stats2")
     message:
          f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - Fasta dir : f"{{input}}"
+|   - Fasta dir : {{input}}
 |-- Output:
-|   - stat file: f"{{output}}"
+|   - stat file: {{output}}
 |-- Others infos :
 |   -LOG error: {{log.error}}
 --------------------
@@ -26,7 +26,7 @@ rule fasta_stats2:
     shell:
         """
         cmd="assembly-stats {params.other_option} {input.fasta_file}";
-        (echo $cmd) > {log.error};
+        (echo $cmd) > {log.output};
         ($cmd) 1>> {output.stat_file} 2>> {log.error};
 
         """   
\ No newline at end of file

frangiPANe/snakefiles/fastq_stats.smk

@@ -4,17 +4,17 @@ rule fastq_stats:
     output:
         stat_file = f"{frangipane_obj.path_stat_fastq}/{{fastq_file}}.stats"
     log:
-        output = f"{frangipane_obj.path_logs}/fastq_stats/{{fastq_file}}.o",
-        error = f"{frangipane_obj.path_logs}/fastq_stats/{{fastq_file}}.e"
+        output = f"{frangipane_obj.path_logs}/stats/{{fastq_file}}.o",
+        error = f"{frangipane_obj.path_logs}/stats/{{fastq_file}}.e"
     message:
          f"""--------------------
--Running {{rule}}
+-Running {rule}
 |-- Input:
-|   - Fastq dir : f"{frangipane_obj.path_fastq}"
+|   - Fastq dir : {frangipane_obj.path_fastq}
 |-- Output:
-|   - Fastq stat dir: f"{frangipane_obj.path_stat_fastq}"
-|-- Others infos :
-|   -LOGs: f"{frangipane_obj.path_logs}/fastq_stats/"
+|   - Fastq stat dir: {frangipane_obj.path_stat_fastq}
+|-- Other infos:
+|   - LOG output : {log.output} 
 --------------------
 """
     singularity:
@@ -25,5 +25,5 @@ rule fastq_stats:
      """
         cmd='fastq-stats -D {input.fastq_file} > {output.stat_file}' ;
         echo $cmd > {log.output};
-        (fastq-stats -D {input.fastq_file}) 1> {output.stat_file} 2>> {log.output}"""
+        (fastq-stats -D {input.fastq_file}) 1> {output.stat_file} 2>> {log.error}"""
 

frangiPANe/snakefiles/filter_length.smk

@@ -4,35 +4,37 @@ rule filter_length:
     output:
         fasta_file = f"{frangipane_obj.path_filtered_contigs}/{{sample_name}}-{{ref_name}}-contigs.fa"
     log:
-        output = f"{frangipane_obj.path_logs}/abyss-filtered/{{sample_name}}-{{ref_name}}-filtered.log.o",
-        error = f"{frangipane_obj.path_logs}/abyss-filtered/{{sample_name}}-{{ref_name}}-filtered.log.e"
+        output = f"{frangipane_obj.path_logs}/post-abyss/filtered-{{sample_name}}-{{ref_name}}.o",
+        error = f"{frangipane_obj.path_logs}/post-abyss/filtered-{{sample_name}}-{{ref_name}}.e"
     params:
-        length = frangipane_obj.get_config_value("CONTIGS","minLength"),
-        dirout = frangipane_obj.path_filtered_contigs
+        length = frangipane_obj.get_config_value("CONTIGS","minLength")
     message:
         f"""--------------------
--Running {{rule}}
+-Running {rule}
 |-- Input:
 |   - fasta files: {frangipane_obj.path_contigs}
 |-- Output:
 |   - fasta dir: {frangipane_obj.path_filtered_contigs}
-|-- Others infos :
-|   -LOG error: 
+|-- Other infos:
+|   - LOG output : {log.output} 
 --------------------
 """
     run:
 
         from Bio import SeqIO
 
-        with open(output.fasta_file, 'w') as o:
-            for seq in SeqIO.parse(input.fasta_file, "fasta"):
-                    if len(seq) >= params.length:
-                        o.write('>' + str(seq.description) + '\n')
-                        txt = str(seq.seq)
-                        lines = txt.split('\n')
-                        for line in lines:
-                            if len(line) > 0:
-                                if line[0] == '>':
-                                    o.write(line + '\n')
-                                else:
-                                    o.write(gs.format_60(line) + '\n')
\ No newline at end of file
+        try:
+            with open(output.fasta_file, 'w') as o:
+                for seq in SeqIO.parse(input.fasta_file, "fasta"):
+                        if len(seq) >= params.length:
+                            o.write('>' + str(seq.description) + '\n')
+                            txt = str(seq.seq)
+                            lines = txt.split('\n')
+                            for line in lines:
+                                if len(line) > 0:
+                                    if line[0] == '>':
+                                        o.write(line + '\n')
+                                    else:
+                                        o.write(gs.format_60(line) + '\n')
+        except Exception as e:
+            raise ValueError(f"Error during filtering on contig legth: {e}")
\ No newline at end of file

frangiPANe/snakefiles/merge_fastq_stats.smk

@@ -10,9 +10,11 @@ rule merge_fastq_stats:
             f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - Fastq stats : f"{frangipane_obj.path_stat_fastq}"
+|   - Fastq stats : {frangipane_obj.path_stat_fastq}
 |-- Output:
-|   - stat file: {{output.stat_file}}
+|   - stat file: {output.stat_file}
+|-- Other infos:
+|   - LOG output : {log.output} 
 --------------------
 """
     run:
@@ -21,6 +23,7 @@ rule merge_fastq_stats:
 
         # open the file merging the fastq-stat results on the whole set of samples
         f = open(output.stat_file, "a+")
+
         header = "sample\tfile\treads\tlen\tlen_mean\tlen_stdev\tlen_min\tphred\twindow-size\tcycle-max\tqual_min\tqual_max\tqual_mean\tqual_stdev\t%A\t%C\t%G\t%T\t%N\ttotal_bases"
         f.write(header + "\n")
 

frangiPANe/snakefiles/merge_flagstat.smk

@@ -13,6 +13,8 @@ rule merge_flagstat:
 |   - Bam stat directory : f"{frangipane_obj.path_bam}"
 |-- Output:
 |   - Stat file: {{output.stat_file}}
+|-- Other infos:
+|   - LOG output : {log.output} 
 --------------------
 """
     run:

frangiPANe/snakefiles/plot_fasta.smk

-## RULE
 rule plot_fasta:
     input:
-        ref=rules.bwa_index.input.ref,
+        ref=rules.bwa_index.input.ref
     output: 
         png = f"{path_dico['plot']}/00_{ref_basename}.png",
         csv = f"{path_dico['sum']}/00_{ref_basename}.txt",
     log:
-        output = f"{frangipane_obj.path_logs}/plot/plot_fasta.o"
-        error = f"{frangipane_obj.path_logs}/plot/plot_fasta.e"
+        output = f"{frangipane_obj.path_logs}/report/plot_fasta.o"
+        error = f"{frangipane_obj.path_logs}/report/plot_fasta.e"
     message:
          f"""--------------------
 -Running {{rule}}
 |-- Input:
-|----- Fasta : {{input.ref}}
-|----- Plot : {{output.png}}
-|----- Csv : {{output.csv}}
+|    - Fasta : {input.ref}
+|-- Output:
+|    - Plot : {output.png}
+|    - Csv : {output.csv}
+|-- Other infos:
+|    - LOG output : {log.output} 
+--------------------
 """
     run:
         gs.dashboard_genome(input.ref, output.png, output.csv)

frangiPANe/snakefiles/plot_fastq.smk

@@ -12,14 +12,17 @@ rule plot_fastq:
     params:
         plot_dir = frangipane_obj.path_stat_plot
     log:
-        output = f"{frangipane_obj.path_logs}/plot/plot_fastq.o",
-        error = f"{frangipane_obj.path_logs}/plot/plot_fastq.e",
+        output = f"{frangipane_obj.path_logs}/report/plot_fastq.o",
+        error = f"{frangipane_obj.path_logs}/report/plot_fastq.e",
     message:
          f"""--------------------
 -Running {{rule}}
 |-- Input:
-|----- Stat file : {{input.csv}}
-|----- Plots : {{params.plot_dir}}
+|    - Stat file : {input.csv}
+|-- Output:
+|    - Plot : {params.plot_dir}
+|-- Other infos:
+|    - LOG output : {log.output} 
 --------------------
 """
     run:

frangiPANe/snakefiles/plot_mapping.smk

@@ -5,13 +5,17 @@ rule plot_mapping:
     output: 
         png_mapping=f"{frangipane_obj.path_stat_plot}/mapping_read_ratio-{frangipane_obj.reference_basename}.png",
     log:
-        output = f"{frangipane_obj.path_logs}/plot/plot_mapping.o",
-        error = f"{frangipane_obj.path_logs}/plot/plot_mapping.e"
+        output = f"{frangipane_obj.path_logs}/report/plot_mapping.o",
+        error = f"{frangipane_obj.path_logs}/report/plot_mapping.e"
     message:
          f"""--------------------
 -Running {{rule}}
 |-- Input:
-|----- Stat : {{input.csv}}
+|    - Stat file : {input.csv}
+|-- Output:
+|    - Plot : {output.png_mapping}
+|-- Other infos:
+|    - LOG output : {log.output} 
 --------------------
 """
     run:

frangiPANe/snakefiles/samtools_flagstat.smk

@@ -5,8 +5,8 @@ rule samtools_flagstat:
     output:
         stat_file = f"{frangipane_obj.path_bam}/{{ref_name}}/{{sample_name}}-{{ref_name}}.flagstat"
     log:
-        output =  f"{frangipane_obj.path_logs}/samtools_flagstat/{{sample_name}}-{{ref_name}}.o",
-        error =  f"{frangipane_obj.path_logs}/samtools_flagstat/{{sample_name}}-{{ref_name}}.e"
+        output =  f"{frangipane_obj.path_logs}/stats/flagstat-{{sample_name}}-{{ref_name}}.o",
+        error =  f"{frangipane_obj.path_logs}/stats/flagstat-{{sample_name}}-{{ref_name}}.e"
     threads: get_threads('samtools_flagstat', 1)
     message:
         f"""--------------------

frangiPANe/snakefiles/samtools_view.smk

@@ -9,20 +9,20 @@ rule samtools_view:
         error =  f"{frangipane_obj.path_logs}/samtools_view/{{sample_name}}-{{ref_name}}.e",
         output = f"{frangipane_obj.path_logs}/samtools_view/{{sample_name}}-{{ref_name}}.o"
     params:
-        other_options = frangipane_obj.get_config_value("PARAMS","samtools_view"), #config['PARAMS']['samtools_view'] if "samtools_view" in config['PARAMS'] else "",
+        other_options = frangipane_obj.get_config_value("PARAMS","samtools_view"),
     threads: get_threads('samtools_view', 1)
     message:
         f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - bam file : {{input.bam_file}}
+|   - bam file : {input.bam_file}
 |-- Output:
-|   - bam file: {{output.bam_file}}
+|   - bam file: {output.bam_file}
 |-- Others infos :
-|   -Threads : {{threads}}
-|   -LOG output: {{log.output}}
-|   -LOG error: {{log.error}}
-|   -Other options: {{params.other_options}}
+|   -Threads : {threads}
+|   -LOG output: {log.output}
+|   -LOG error: {log.error}
+|   -Other options: {params.other_options}
 --------------------
 """
     singularity:
@@ -35,6 +35,6 @@ rule samtools_view:
         (echo $cmd) > {log.output};
         ($cmd) 1>> {log.output} 2>> {log.error};
         
-        cmd="samtools index -@ {threads} {output.bam_file}";
-        (echo $cmd) >> {log.output};
-        ($cmd) 1>> {log.output} 2>> {log.error}"""
\ No newline at end of file
+        cmd2="samtools index -@ {threads} {output.bam_file}";
+        (echo $cmd2) >> {log.output};
+        ($cmd2) 1>> {log.output} 2>> {log.error}"""
\ No newline at end of file

frangiPANe/snakefiles/vecscreen.smk

@@ -8,16 +8,16 @@ rule vecscreen:
         error =  f"{frangipane_obj.path_logs}/vecscreen/{{sample_name}}-{{ref_name}}.log.e",
         output = f"{frangipane_obj.path_logs}/vecscreen/{{sample_name}}-{{ref_name}}.log.o"
     params:
-        length = frangipane_obj.get_config_value('CONTIGS','minLength') #config['CONTIGS']['minLength']
+        length = frangipane_obj.get_config_value('CONTIGS','minLength')
     message:
         f"""--------------------
 -Running {{rule}}
 |-- Input:
-|   - fasta files: {{input.fasta_file}}
+|   - fasta files: {input.fasta_file}
 |-- Output:
-|   - fasta dir: {{output.fasta_file}}
+|   - fasta dir: {output.fasta_file}
 |-- Others infos :
-|   - Filtering on sequences longer than  : {{params.length}}
+|   - Filtering on sequences longer than  : {params.length}
 --------------------
 """
     run: