Good Morning Gordon: Create your data in bedGraph four-column format. Use the bigWig encoder to encode that into a wiggle graph. This will be the most efficient encoding of this type of data. See also: http://genomewiki.ucsc.edu/index.php/Selecting_a_graphing_track_data_format
The bedGraph four-column specification is equivalent to having different "spans" of data sizes. In the older wiggle encoding system this was highly inefficient under the hood because each line in the bedGraph was turned into fixedStep span=1 data points and it blows up with a tremendous amount of data. The bigWig encoder overcomes this limitation. You can not mix "spans" of different sizes. You can not have overlapping data. The bigWig encoder will detect that condition and refuse to encode the data. --Hiram Gordon Robertson wrote: > Hello, > > I need to display genome-scale profiles in mm9. The profiles will consist of > long sections with constant heights that are separated by shorter sections > in which heights vary. > > Because the files are large, I'd prefer to use the compact fixedStep format. > A year ago I did such work with files that combined fixedStep sections and > BED-format lines, but the WIG format description has changed since then. > > Now, your help page (genome.ucsc.edu/goldenPath/help/wiggle.html) says: > "Note that for both variableStep and fixedStep formats, the same span must > be used throughout the dataset. If no span is specified, the default span of > one is used." > > The span restriction seems to require me to use either: a) fixedStep > inefficiently (e.g. span=1, even through kb-long constant-height sections), > or b) bedGraph, which would give large files for which a bigBedGraph format > is unavailable (or does bigBed also handle bedGraph?). > > For this application, it will probably be practical to use a fixedStep WIG > after converting the file to bigWig. But do I understand the span > restriction correctly? I ask because the WIG format design is generally very > simple and effective, and it's less obvious why span values would be > restricted. > > Could you clarify two other points that are not stated explicitly on your > WIG help page? > 1. Can a single WIG file contain only fixedStep or variableStep sections, > but not both types? > 2. Do the current fixedStep and variableStep formats permit WIG 'blocks' to > overlap? Overlapped blocks could be generated by values assigned to start, > step and span. > > Thanks for your help with these points. > > G _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
