Re: [Gmod-ajax] Question about exact align between different tracks at high resolution

[Mitch S. <mit...@be...>]

lfgu wrote:
> I will take your test dataset to descript this question.
>  
> The exact region for the second exon of rna-Apple3 genes in your volvox.gff3 file is 19000-19500.
>  
> When you press the zoom in button until you can see bases at DNA tracks,then you will find that the Figure for second exon structure in 
> Exonerate predictions tracks is not located from 19000 to 19500. 
>  
> The align between  DNA Tracks and Exonerate predictions tracks is not exact align at single base resolution, 
> there are several bases deviation.
>  
> I also check many cases on my own dataset, it is a pervasive phenomenon.

There are a few things going on here.  One is that some web browsers are 
less accurate than others; ticket #12 is for the JBrowse changes 
required to work around this, which I plan to work on soon (once the 
track density scalability work is done).  Firefox seems more accurate 
than other browsers, so right now you can get somewhat better results 
using that.  What browser are you using?

Another issue is one-based vs. zero-based; we haven't really talked 
about this, so maybe it's a good time to address it.  Currently, JBrowse 
uses interbase coordinates (also known as zero-based, half-open 
coordinates).  Interbase is what the UCSC browser, the chado database, 
and the BED file format use.  The GBrowse genome browser and the GFF 
file format use 1-based, fully closed coordinates.  So the feature that 
starts at "19000" in the GFF file starts one base before "19000" in JBrowse.

For comparison, I've attached screenshots the beginning of a particular 
exon in JBrowse, UCSC, and GBrowse (modENCODE) (the transcript is 
CG42358-RA).  While the numbering is different between the 1-based 
browser and the 0-based browsers, you can see that the DNA sequence and 
the exon boundary line up the same in all of them (with the exon 
starting at the cytosine).

Regards,
Mitch

JBrowse:


UCSC:


GBrowse: