How Are Features With No Strand Encoded In Ucsc Bed Format?

Question

[Casey Bergman]

Hi -

I have a set of annotations that have strand information for some of them, but no strand information for others. I'd like to visualise them in the UCSC Genome Browser in BED format, but the UCSC FAQ on BED format only specifies strand as being "+" or "-". Is anyone aware of documentation about how UCSC BED format handles ambigous stranded features?

Many thanks, Casey

Answer 1

[Casey Bergman]

I cannot find any documentation on this but empirically, it seems that encoding unknown strand as "." works in the Genome Browser. Features with strand set to "." visualize properly but have no >>> or <<< arrowheads in the browser. Also, BED files with "." in the strand column are read properly by UCSC source tools such as bedSort and bedIntersect.

[Istvan Albert]

Yes the dots are the way to do this (though I can't find a documentation for this either). This applies to all other fields as well (other than chrom, start, end) moreover the same is true for GFF and VCF files too. At least the GFF3 spec states that:

Undefined fields are replaced with the "." character, as described in the original GFF spec

Edit: one important rationale for adding dots compared to just leaving it empty is that the default white space split in most programming languages (where the programmer does not specify the splitting character) will treat consecutive white space characters as a single separator thus shifting columns.