Chromosome classification result about genomad

[apcamargo]

This is a good question, but one without a clear-cut answer. The classifier works by placing sequences on a continuum, and there's no true "unclassified" category if we consider that a sequence should be assigned to the class with the highest score out of the three. I've set the minimum score to 0.7 for plasmids and viruses because users are typically interested in the most confident classifications, and 0.7 has proven to be a good cutoff empirically. However, it's not as simple as saying that a sequence with a score of 0.7 should be confidently classified as a plasmid while one with a score of 0.6999 should be left "unclassified."

If you're interested in the plasmid/chromosome ratio, I suggest using the highest score from the <prefix>_aggregated_classification.tsv file (if you're working with metagenomes or many isolate genomes, you might benefit from the score calibration feature). If you actually want to set apart a set of sequences to label as "plasmids," you may consider the score (as a proxy for the reliability of the classification) as well as other features, such as the number of hallmarks.

[LeabaeL]

Thanks for your reply, I will define it later and see how robust the results are based on the highest score statistic.