Crude AMR database harmonisation

We need a quick way to compare results between different AMR databases While there are better ideas of dealing with this problem more robustly (...speak to me about genomically-informed nomenclature normalisation), for mapping NCBI and ResFinder to CARD's ARO:

• Download NCBI and ResFinder databases

• Run them through RGI

• Using hits, create a mapping of NCBI and ResFinder genes to the ARO

• Highlight any situation of no hits/disconnect for manual curation

Running

conda env create -f env.yml
conda activate crude_harmonisation
bash crude_db_harmonisation.sh

Output mapping can be found at resfinder_ncbi_ARO_mapping.tsv

• 93.1% of ResFinder entries with trivial mapping to ARO using homology models (218 entries missing).
• 80.7% of NCBI entries with trivial mapping to ARO using homology models (1260 entries missing).