Compute Cluster Signatures¶
p3-signature-clusters < peg.data > cluster.signatures
The standard input file normally contains [n/a, n/a, family_id, feature_id, contig, start, end, strand, function](the first two columns are ignored). Such a file is created from a family signatures file (output of p3-signature-familes) using the following command
p3-signature-peginfo --gs1=FileOfGenomeIds < family.data > peg.data
However, any file containing the appropriately named columns in the headers (see below) will work.
This script produces a file containing entries of the form
famId1 peg1 func1 famId2 peg2 func2 . . . //
if the –verbose flag is used. Else, you get 1 line per cluster of the form
genome1 peg11,peg12,...,peg1n1 genome1 peg13,peg14,...,peg1n2 genome2 peg21,peg22,...,peg2n1 . . .
that is, you get the genome containing the cluster followed by a comma-separated list of peg ids. There is often a number of clusters for a single genome.
In the non-verbose mode you get a header line containing
There are no positional parameters.
The standard input is specified using the options in Input Options. It should be a tab-delimited file with headers, containing the following fields at minimum.
The ID of a protein family.
The ID of a feature in the protein family.
The ID of the contig containing the feature.
The index of the leftmost location for the feature on the contig.
The index of the rightmost location for the feature on the contig.
The strand containing the feature (
The function assigned to the feature.
The additional command-line options are as follows.
In normal mode, clusters are written in a readable format, and the family id and the peg function are included for each member of a cluster. In terse mode, each cluster is written on a single line.
Maximum base-pair distance between the midpoints of two features in order for them to be considered close. The default is 2000.