This is an inconsistent error because it is typically right at the edge of memory limits and background processes like docker can tip it over the top. The HCMI omics script downloads hundreds (thousands?) of samples and creates a dataframe for each before concatenating. By streaming each to file, this would avoid holding all of them in memory at once and causing a crash.

Memory error:

build_omics.sh: line 7:     7 Killed                  python 02-getHCMIData.py -m full_manifest.txt -t transcriptomics -o /tmp/hcmi_transcriptomics.csv.gz -g $1 -s $2
Error on or near line 7 while executing: python 02-getHCMIData.py -m full_manifest.txt -t transcriptomics -o /tmp/hcmi_transcriptomics.csv.gz -g $1 -s $2