bundlelinks - bundle neighbouring links together in an effort to reduce complexity of link structure
bundlelinks -links linkfile.txt -max_gap NUM
{-min_bundle_membership NUM | -strict }
{-min_bundle_size NUM}
{-min_bundle_identity FRACTION}
bundlelinks -links data/dog.vs.human.all.txt > data/dog.vs.human.bundles.txt
The purpose of this script is to turn a large set of links into a smaller one by merging neighbouring links.
The script produces a new link file to STDOUT.
A tally is sent to STDERR that lists the following
# total number of links read in num_links 39978 # number of links in initial bundles (filtered for membership) num_links_in_init_bundles 39839 # total number of initial bundles num_init_bundles 524 # number of accepted bundles (filtered for both size and membership) num_passed_bundles 277 (52.86%) # number of links in accepted bundles num_links_in_passed_bundles 36732 (92.20%)
Adjacent links are merged into bundles if their start/end coordinates are sufficiently close. Given two links L1 and L2, they are merged into a bundle if
chr( start(L1) ) == chr( start(L2) ) chr( end(L1) ) == chr( end(L2) )
distance( start(L1), start(L2) ) <= MAX_GAP distance( end(L1), end(L2) ) <= MAX_GAP
If a link does not have any acceptable adjacent neighbouring links, it forms a single-link bundle.
These parameters filter bundles with few links. You can set the minimum number of links required in a bundle for the bundle to be accepted (-min_bundle_membership).
The -strict option is equivalent to -min_bundle_membership 2.
In addition to, or in place of, filtering bundles based on the number of links they comprise, you can accept bundles based on the size of the links in the bundle.
The minimum size parameter is applied independently to both ends of all links in a bundle.
This parameter filters bundles based on the bundle identity, which is defined as
identity = size(merged links) / extent(merged links)
Both ends of the bundle are evaluated independently.
Started and versioned.
Martin Krzywinski
Martin Krzywinski Genome Sciences Centre Vancouver BC Canada www.bcgsc.ca martink@bcgsc.ca