Repository navigation
5.12.1
-
Behavior fix:
gcluster.run()jobs run against the caller's database, working directory, datasets and virtual tracks, and load misha from the caller's library. They had used the example database unless the job calledgsetroot()itself. -
Behavior fix:
gtrack.liftover()could read an indexed source track from the wrong chromosomes, with no error: any 2D track, and 1D tracks in a per-chromosome database whosechrom_sizes.txtnames lack the "chr" prefix. Re-lift such tracks. An indexed source track outside a database is now an error. -
Behavior fix:
gtrack.liftover()of a 2D points track (e.g. Hi-C) writes a points track instead of 1bp rectangles, with memory bounded per chromosome pair (gmax.data.size), so genome-scale Hi-C tracks can be lifted. -
Behavior fix:
gtrack.2d.import()andgtrack.2d.import_contacts()reject a negativestart2. Such rows were silently dropped or moved. -
gtrack.2d.import()andgtrack.2d.import_contacts()run in parallel (up togmax.processesworkers) with identical output. A chromosome pair starts only when its estimated memory fits withingmax.mem.usage; its default is 80% of the machine's RAM, so on a cluster set it to the job's share.gmultitasking = FALSErestores the serial import. -
The 2D imports no longer crash R on an input file they cannot open or with
gmax.data.sizebelow 1. -
An integer
gmax.mem.usageabove about 2 GB (e.g.4000000L) overflowed; it is now read correctly.