Metagenomic AMR gene detection using hierarchical machine learning models based on either AMR curated gene families or sequence identity based clusters.
- Initial read analysis using heuristically-accelerated homology searches (DIAMOND)
- Xgboost based classification of metagenomics reads to AMR gene families or sequence clusters
- Xgboost based classification of reads within family/clusters to specific genes
- Localised assembly of classified reads
- Attempted extension of family/cluster member-specific contigs
External dependencies:
- DIAMOND
- vsearch (cd-hit actually used)
- art
- scikit-learn
- biopython
- tqdm