Best Approach to Construct a Phylogenetic Tree from GFA Files (Within-Species Breeds)

[BlackSlipper]

Hello,

I’m interested in constructing a phylogenetic tree for different breeds within a species using GFA files generated from a cactus-minigraph pangenome pipeline.

I've been informed that it might be possible to build a distance matrix of multiple breeds using panacus software from

ComparativeGenomicsToolkit/cactus#1722 (comment)

However, I wasn't able to find a command line to execute computing distance matrix.

Could you please provide some guidance or point me in the right direction?

Thank you in advance for your help!

[danydoerr]

Hi,

panacus cannot compute a phylogenetic distance matrix. It can produce a similarity matrix (see help output below), from which you can derive dissimilarities that have no metric guarantees. However, I'd suggest a more traditional approach, using standard software such as RAxML on a multiple sequence alignment derived from the pangenome graph, or, if you want to have a quick solution, use an alignment free approach such as SANS on the DNA sequences extracted from the haplotype paths.

Compute coverage table for count type

Usage: panacus similarity [OPTIONS] <GFA_FILE>

Arguments:
  <GFA_FILE>  graph in GFA1 format, accepts also compressed (.gz) file

Options:
  -s, --subset <FILE>            Produce counts by subsetting the graph to a given list of paths (1-column list) or path coordinates (3- or 12-column BED file)
  -e, --exclude <FILE>           Exclude bp/node/edge in growth count that intersect with paths (1-column list) or path coordinates (3- or 12-column BED-file) provided by the given file; all intersecting bp/node/edge will be exluded also in other paths not part of the given list
  -g, --groupby <FILE>           Merge counts from paths by path-group mapping from given tab-separated two-column file
  -H, --groupby-haplotype        Merge counts from paths belonging to same haplotype
  -S, --groupby-sample           Merge counts from paths belonging to same sample
  -a, --total                    Summarize by totaling presence/absence over all groups
  -c, --count <count>            Graph quantity to be counted [default: node] [possible values: node, bp, edge]
  -m, --method <cluster_method>  Method for clustering results [default: centroid] [possible values: single, complete, average, weighted, ward, centroid, median]
  -t <COUNT>                     Set the number of threads used (default: use all threads) [default: 0]
  -v, --verbose                  Set the number of threads used (default: use all threads)
  -h, --help                     Print help```