unclustered_recruitment.py
usage: unclustered_recruitment.py
Recruit unclustered contigs given metagenome annotations and Autometa binning
results. Note: All tables must contain a 'contig' column to be used as the
unique table index
optional arguments:
-h, --help show this help message and exit
--kmers KMERS Path to normalized kmer frequencies table. (default:
None)
--coverage COVERAGE Path to coverage table. (default: None)
--binning BINNING Path to autometa binning output [will look for
col='cluster'] (default: None)
--markers MARKERS Path to domain-specific markers table. (default: None)
--output-binning OUTPUT_BINNING
Path to output unclustered recruitment table.
(default: None)
--output-main OUTPUT_MAIN
Path to write Autometa main table used during/after
unclustered recruitment. (default: None)
--output-features OUTPUT_FEATURES
Path to write Autometa features table used during
unclustered recruitment. (default: None)
--taxonomy TAXONOMY Path to taxonomy table. (default: None)
--taxa-dimensions TAXA_DIMENSIONS
Num of dimensions to reduce taxonomy encodings
(default: None)
--additional-features [ADDITIONAL_FEATURES ...]
Path to additional features with which to add to
classifier training data. (default: [])
--confidence CONFIDENCE
Percent confidence to allow classification (confidence
= num. consistent predictions/num. classifications)
(default: 1.0)
--num-classifications NUM_CLASSIFICATIONS
Num classifications for predicting/validating contig
cluster recruitment (default: 10)
--classifier {decision_tree,random_forest}
classifier to use for recruitment of contigs (default:
decision_tree)
--kmer-dimensions KMER_DIMENSIONS
Num of dimensions to reduce normalized k-mer
frequencies (default: 50)
--seed SEED Seed to use for RandomState when initializing
classifiers. (default: 42)