Merge pull request #47 from pinellolab/fix-onlyref-search

Fix onlyref search
pinellolab · Feb 1, 2024 · b2c6679 · b2c6679
2 parents 0c7ed90 + 7f1ded3
commit b2c6679
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 19 deletions.
diff --git a/PostProcess/post_analisi_snp.sh b/PostProcess/post_analisi_snp.sh
@@ -23,14 +23,15 @@ final_res_alt=${12}
 
 key=${13}
 
-if ! [ -f "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt" ]; then
-	touch "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt"
-fi
 
 echo "Processing SNPs for $key"
 LC_ALL=C grep -F -w $key "$output_folder/crispritz_targets/${ref_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt" >"$output_folder/crispritz_targets/${ref_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key"
-LC_ALL=C grep -F -w $key "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt" >"$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key"
-./scriptAnalisiNNN_v3.sh "$output_folder/crispritz_targets/${ref_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key" "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key" "$output_folder/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${annotation_name}_${mm}_${bDNA}_${bRNA}_$key" "$annotation_file" "${dict_folder}/my_dict_${key}.json" "${ref_folder}/${key}.fa" $mm $bDNA $bRNA "$guide_file" "$pam_file" "$output_folder" || {
+if ! [ -f "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt" ]; then
+	cp "$output_folder/crispritz_targets/${ref_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key" "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key"
+else
+	LC_ALL=C grep -F -w $key "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt" >"$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.$key"
+fi
+./scriptAnalisiNNN_v3.sh "$output_folder/crispritz_targets/${ref_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.${key}" "$output_folder/crispritz_targets/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${mm}_${bDNA}_${bRNA}.targets.txt.${key}" "$output_folder/${ref_name}+${vcf_name}_${pam_name}_${guide_name}_${annotation_name}_${mm}_${bDNA}_${bRNA}_$key" "$annotation_file" "${dict_folder}/my_dict_${key}.json" "${ref_folder}/${key}.fa" $mm $bDNA $bRNA "$guide_file" "$pam_file" "$output_folder" || {
 	echo "CRISPRme ERROR: SNP analysis failed (script: ${0} line $((LINENO-1)))" >&2
 	exit 1
 }

diff --git a/PostProcess/submit_job_automated_new_multiple_vcfs.sh b/PostProcess/submit_job_automated_new_multiple_vcfs.sh
@@ -54,7 +54,7 @@ dummy_file="${output_folder}/.dummy.txt"
 touch "${output_folder}/.empty.txt"
 empty_file="${output_folder}/.empty.txt"
 ##CREATE EMPTY DIR
-mkdir "${output_folder}/.empty"
+mkdir -p "${output_folder}/.empty"
 empty_dir="${output_folder}/.empty"
 
 rm -f $output_folder/queue.txt
@@ -548,21 +548,15 @@ sampleID=$output_folder/.sampleID.txt
 # echo -e 'Merging targets' >  $output
 
 #create result file for each scoring method
-# echo "header" >$final_res.bestCFD.txt
-# echo "header" >$final_res.bestmmblg.txt
-# echo "header" >$final_res.bestCRISTA.txt
-
+header="#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref"
 #header into final_res best
-sed -i '1i #Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestCFD.txt"
-sed -i '1i #Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestmmblg.txt"
-sed -i '1i #Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestCRISTA.txt"
+sed -i '1 i\#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestCFD.txt"
+sed -i '1 i\#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestmmblg.txt"
+sed -i '1 i\#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref' "$final_res.bestCRISTA.txt"
 #header into final_res alt
-echo "header" >$final_res_alt.bestCFD.txt
-echo "header" >$final_res_alt.bestmmblg.txt
-echo "header" >$final_res_alt.bestCRISTA.txt
-sed -i '1 s/^.*$/#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref/' "$final_res_alt.bestCFD.txt"
-sed -i '1 s/^.*$/#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref/' "$final_res_alt.bestmmblg.txt"
-sed -i '1 s/^.*$/#Bulge_type\tcrRNA\tDNA\tReference\tChromosome\tPosition\tCluster_Position\tDirection\tMismatches\tBulge_Size\tTotal\tPAM_gen\tVar_uniq\tSamples\tAnnotation_Type\tReal_Guide\trsID\tAF\tSNP\t#Seq_in_cluster\tCFD\tCFD_ref/' "$final_res_alt.bestCRISTA.txt"
+printf $header >$final_res_alt.bestCFD.txt
+printf $header >$final_res_alt.bestmmblg.txt
+printf $header >$final_res_alt.bestCRISTA.txt
 
 echo -e 'Merging Targets\tStart\t'$(date) >>$log
 #SORT FILE TO HAVE CHR AND POS IN PROXIMITY TO MERGE THEM