Good Afternoon Brian:

Do I understand that you originally had bedGraph files which you converted
to wiggle files via wigEncode ?  If that is true, you may have constructed
much larger files in the conversion than was originally in the bedGraph
files.  That is not necessarily true though, it depends upon the bedGraph
data and how it was organized.  The .wib files are about as condensed as
you can get for this type of data with a single byte for each data value
in the graph.  However, a bedGraph data value, for example:

chr1  0   1000  3.14

converts into 1,000 individual data values when converted to wiggle data.
If you had the bedGraph files, you could use them directly for the source
of the graph and in this example, 16 bytes in the bedGraph file vs. over 1,000
bytes in the wiggle representation.  However, if you had 1,000 different
data values for the positions in chr1 from 0 to 1000, the .wib representation
would remain at 1000 bytes and the bedGraph would explode to more than 16000 
bytes
for the 1,000 lines of data.  A full genome sized data set in .wib 
representation
would always be less than 3 Gb.  bedGraph files would be immense with 3G lines
of ascii data.

You can not return to the original data values from the wiggle compressed
single bytes.  The compression changes the data which is fine for the
graph resolution, but it isn't the same data.  Please note this discussion
of data types:
http://genomewiki.ucsc.edu/index.php/Selecting_a_graphing_track_data_format

If you have your original bedGraph files and they are smaller than the .wib
files, you could use your bedGraph files.

--Hiram

----- Original Message -----
From: "Brian J. Abraham" <[email protected]>
To: [email protected]
Sent: Friday, December 3, 2010 2:28:59 PM GMT -08:00 US/Canada Pacific
Subject: [Genome] WIB contents and conversion into bedGraph

Hello,

I am running a local mirror of the UCSC browser and am coming very
close to reaching storage capacity.  We went through a phase where we
uploaded local tracks in the .wig / .wib format by creating these
files via wigEncode.  We have come to conclusion that the > 1GB per
track cost is too high for the resolution and our use.  Is there a way
to reconstitute bedGraphs directly from the .wig / .wib files or from
within MySQL?  Are we taking storage hits for both the .wib file and
what is contained in the database?

Many thanks,
~Brian J. Abraham
_______________________________________________
Genome maillist  -  [email protected]
https://lists.soe.ucsc.edu/mailman/listinfo/genome

Reply via email to