Fixes from prod (#231)

* Allow more characters in keys/values of novel attribute
cast subject as string

* For a single process to run accessioning

* Specify the NCBI key to enable more request to NCBI API

eva_submission/eload_ingestion.py

@@ -166,7 +166,7 @@ def get_db_name(self, assembly_accession):
         """
         # query EVAPRO for db name based on taxonomy id and accession
         with self.metadata_connection_handle as conn:
-            db_name = resolve_variant_warehouse_db_name(conn, assembly_accession, self.taxonomy)
+            db_name = resolve_variant_warehouse_db_name(conn, assembly_accession, self.taxonomy, ncbi_api_key=cfg.get('eutils_api_key'))
             if not db_name:
                 raise ValueError(f'Database name for taxid:{self.taxonomy} and assembly {assembly_accession} '
                                  f'could not be retrieved or constructed')

eva_submission/etc/eva_project_conf.yaml

@@ -121,6 +121,7 @@ Sample:
     Title: string
     cultivar: string
     Description: string
+    Subject: string
 
 Files:
   required:

eva_submission/etc/eva_project_validation.yaml

@@ -106,7 +106,7 @@ Sample:
      cell_line: {           type: 'string',  nullable: True}
      serotype: {            type: 'string',  nullable: True}
      serovar: {             type: 'string',  nullable: True}
-     Novel attribute(s): {  type: 'string',  nullable: True, regex: '[\w.\-]+:[\w.\-]+(,\s*[\w.\-]+:[\w.\-]+)*'}
+     Novel attribute(s): {  type: 'string',  nullable: True, regex: '[\w.\-_ ]+:[\w.\-_ ]+(,\s*[\w.\-_ ]+:[\w.\-_ ]+)*'}
 
 Files:
   type: 'list'

eva_submission/nextflow/accession_and_load.nf

@@ -214,6 +214,8 @@ process accession_vcf {
     clusterOptions "-o $params.logs_dir/${log_filename}.log \
                     -e $params.logs_dir/${log_filename}.err"
 
+    maxForks 1
+
     input:
     tuple val(vcf_filename), val(vcf_file), val(assembly_accession), val(aggregation), val(fasta), val(report)