Merge pull request #56 from EBI-Metagenomics/dev

Prevent generate_extra_weight_table.py from bailing out if a genome is missing.

bin/generate_extra_weight_table.py

@@ -209,10 +209,10 @@ def add_genome_info(genome_info_file, extra_weights):
  if genome in extra_weights:
  extra_weights[genome] = weight
  else:
- sys.exit(
+ logging.warning(
  "Extra weight information for genome {} was provided but genome is"
- " not found in the genomes folder. Check naming format - is the"
- " extension missing? Extra weight information cannot be used.".format(
+ " not found in the genomes folder. Genome possibly removed during the QC step."
+ " Extra weight information cannot be used.".format(
  genome
  )
  )

helpers/file_organiser.sh

@@ -95,6 +95,8 @@ function CopyAdditionalFiles {
  cp -r additional_data ${SAVE_TO_PATH}/${CATALOGUE_FOLDER}/${CATALOGUE_VERSION}/
  cd ${SAVE_TO_PATH}/${CATALOGUE_FOLDER}/${CATALOGUE_VERSION}/additional_data
  bsub -q production -M 1G -n 1 -o /dev/null "tar -czvf mgyg_genomes.tar.gz mgyg_genomes && rm -r mgyg_genomes"
+ cd ${SAVE_TO_PATH}/${CATALOGUE_FOLDER}/${CATALOGUE_VERSION}/gene_catalogue
+ bsub -q production -M 1G -n 1 -o /dev/null "gzip gene_catalogue-100.ffn"
 }