-
Notifications
You must be signed in to change notification settings - Fork 3
FAQ
Put the new files in the input folder and run the same command with the same output folder. Existing sketches are reused, so only the new genomes are sketched. Distances and trees are recomputed for all genomes.
Yes, as long as each sample is either assemblies or reads. Read sketches can include sequencing errors: use
--min-copies (default 2) to filter k-mers seen only once, and expect distances involving low-coverage samples to be
slightly larger.
From the file names, without the extension; read suffixes such as _R1/_R2 are also removed from fastq files.
See Usage.
Check sample_stats.tsv: a failed sample could not be read by Mash (empty, truncated or not a fasta/fastq file).
Also check the log for "Ambiguous sample names".
They are unrooted. Root them in your tree viewer on an outgroup (a genome from a related species) or on the midpoint.
The UPGMA tree (_hc.nwk) is already rooted.
Usually because the genomes are very closely related: their sketches differ by very few hashes. See Bootstrap support.
ANI is roughly 1 − Mash distance for related genomes. For accurate ANI values, use skani or FastANI (see Related tools).
The defaults (k = 21, 10,000 hashes) work well for bacterial genomes. A larger sketch gives more precise distances between very close genomes. See Usage.
Yes: dendrogram-from-matrix builds the same trees and PCoA from any square distance matrix
(see Other tools).
Yes: --pcoa --metadata metadata.tsv --color-by COLUMN. The first column of the metadata file must hold the sample
names. Each group gets its own colourblind-friendly colour and marker shape.
Mash works with any genomes, but genome_comparator has mostly been used and tested on bacterial genomes.
Getting started
Understanding the results
Reference