attempt to download missing files

src/Dockerfile

@@ -19,7 +19,7 @@ ARG AWS_SECRET_ACCESS_KEY
 #ENV AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY
 
 #SRA toolkit (from serratus-dl)
-ENV SRATOOLKITVERSION='2.10.4'
+ENV SRATOOLKITVERSION='2.10.7'
 RUN wget --quiet https://ftp-trace.ncbi.nlm.nih.gov/sra/sdk/${SRATOOLKITVERSION}/sratoolkit.${SRATOOLKITVERSION}-centos_linux64.tar.gz &&\
     tar xzf sratoolkit.${SRATOOLKITVERSION}-centos_linux64.tar.gz &&\
     rm -f sratoolkit.${SRATOOLKITVERSION}-centos_linux64.tar.gz &&\
@@ -45,6 +45,8 @@ RUN chmod +x parallel-fastq-dump
 RUN wget --quiet http://opengene.org/fastp/fastp 
 RUN chmod +x fastp 
 
+RUN mkdir -p /mnt/serratus-data # for local testing
+
 RUN pwd
 RUN df -T
 RUN ls

src/batch_processor.py

@@ -80,7 +80,7 @@ def process_file(accession, region):
         # 'is larger than maximum allowed: skipped' solved by:
         # https://wiki.rc.hms.harvard.edu/display/O2/Aspera+to+download+NCBI+SRA+data
         #os.system('prefetch --max-size 35G '+accession)
-        os.system('prefetch '+accession)
+        os.system('prefetch --max-size 35G '+accession)
         prefetch_time = datetime.now() - prefetch_start 
         sdb_log(sdb,accession,'prefetch_time',int(prefetch_time.seconds))
         os.system('mkdir -p tmp/')