remove index from pipeline, add error and readme section

README.md

@@ -63,6 +63,7 @@ If you want to try other viruses, or use your own reference and bed files:
 
 - Look at our directory "DENV_primers_and_refs" for formatting file names etc
 - Provide the stem of each file in the "refs.txt" tesxt file in the same folder
+- Make sure you have indexed your references by typing "bwa index <reference.fasta>"
 - Use the ``--reference-directory`` option to provide the path to the directory. 
 
 

denv_pipeline/scripts/mapper.sh

@@ -16,11 +16,6 @@ while IFS= read -r virustype || [[ -n "$virustype" ]]; do
     trimbed=${primer_dir}/${virustype}.trim.bed
     consensus_name=${fname}.${virustype}
 
-    if ! [ -f ${primer_dir}/${virustype}.fasta.ann ]; then
-        echo "----->>>> making index files for ${virustype}"
-        bwa index ${fasta}
-    fi
-
     echo "----->>>>>Mapping reads against serotype "${virustype}" reference sequence"
     which bwa
     bwa mem -v 1 -t 2 -O 10 ${fasta} $read1 $read2 | samtools view -bS -F 4 -F 2048 | samtools sort -o ${tempdir}/${fname}.${virustype}.bam

denv_pipeline/utils/error_checks.py

@@ -92,13 +92,17 @@ def check_primer_dir(config):
 
                 bed_file = f'{virus_type}.bed'
                 fasta_file = f'{virus_type}.fasta'
+                index_file = f'{virus_type}.fasta.amb'
 
                 if not bed_file in all_files:
                     sys.stderr.write(green(f"Error: Missing bed file for {virus_type} in {config['reference_directory']}. I am expecting it to be called {bed_file} and the '.bed' has to be lower case.\n"))
                     sys.exit(-1)
                 if not fasta_file in all_files:
                     sys.stderr.write(green(f"Error: Missing reference file for {virus_type} in {config['reference_directory']}. I am expecting it to be called {fasta_file}\n"))
                     sys.exit(-1)
+                if not index_file in all_files:
+                    sys.stderr.write(green(f"Error: Missing reference index files for {virus_type} in {config['reference_directory']}. I am expecting it to be called {index_file}\n To generate this, please use bwa index <reference sequence>.\n"))
+                    sys.exit(-1)
 
                 with open(os.path.join(config["reference_directory"],fasta_file)) as f:
                     seq_count = 0