[Genome] WIG display controls: y-axis scaling

[Hiram Clawson]

Good Afternoon Gordon:

Thanks for the input.  We can certainly consider allowing minimum
to be fixed, and allow the maximum to auto-scale.  That is certainly
reasonable.

To get around the puzzling behavior you mention, use the viewLimits
option on your track line, for example viewLimits=0:10 will allow the
settable minimum and maximum to be 0 to 10 instead of the discovered
minimum and maximum that is found in the data when no viewLimits
are specified.

See also other options for your track line:
http://genome.ucsc.edu/goldenPath/help/wiggle.html

Be especially wary of the windowingFunction option.  This can have a dramatic
effect on the graphed data when large sections of a chromosome are in view
and a lot of data points need to be combined into a single pixel graph.  For
example, on the GC track, the windowing function is "mean" and when you
view a whole chromosome it looks like all the data is between 33 and 60
for a chr1 view on Human Mar 2006.  This is because over 80,000 individual
data points are being averaged together to get a single pixel drawn on
the graph.  At such a view, "maximum" shows everything is at 100, and
"minimum" shows everything at zero.

--Hiram

Gordon Robertson wrote:
> Hello
> 
> Could I ask whether I understand two aspects of the y-axis 'view scaling'
> controls for WIG custom tracks: 1) I do not see a behaviour that I would
> like to have available, and 2) I see a puzzling behaviour.
> 
> The application is to display actual and test/synthetic ChIP-Seq peak sets
> (human and mouse). 
> 
> 1. I find two 'data view scaling' behaviours: 'auto-scale to data view' and
> 'use vertical viewing range setting'.
> 
> I'd prefer to be able to fix the minimum y-axis value at 0 or 1 and to leave
> the maximum value to autoscale to the data. This behaviour would facilitate
> working flexibly with ChIP-Seq WIG files, over a wide range of display
> region lengths. Currently, on 'auto-scale', the minimum peak height
> displayed seems to be the minimum value in a displayed peak, and peaks that
> have a minimum value greater than 1 often need to be toggled manually to a
> fixed viewing range setting in order to display the full peak data range
> (but see 2, below). However, a fixed range setting is a poor fit to quickly
> surveying large genomic regions while seeing maximum detail and full dynamic
> range in every view.
> 
> The view scaling behaviour appears to override the ' Draw indicator line at
> y = [  ]' setting. That is, I cannot force the minimum autoscale value to 0
> by toggling the indicator line 'on'.
> 
> 2. There is a puzzling aspect to the behaviour. Submitting the example WIG
> file will give a display with a minimum (auto-scaled) value of 2, and I
> cannot manually set the minimum y axis range value to less than 2, though
> I'd prefer to have a 1 or 0 available as a minimum.
> 
> If I understand the controls correctly, could I ask that you consider
> creating a third data view scaling choice that would let a user fix a lower
> display value but let the maximum autoscale?
> 
> Thank you for your help.
> 
> A small WIG file that illustrates this behaviour is:
> 
> browser position chr13:26133900-26133925
> track name="tc" description="test case" type="wiggle_0" color=50,50,150
> yLineMark=0.0 yLineOnOff=on visibility=2 maxHeightPixels=40:40:2 priority=1
> fixedStep chrom=chr13 start=26133907 step=1
> 2
> 2
> 2
> 2
> 3
> 4
> 9
> 4
> 3
> 3
> 2
> 2
> 2
> 
> 
> G 
> ---
> Gordon Robertson
> Gene Regulation Informatics
> Canada's Michael Smith Genome Sciences Centre
> Vancouver BC Canada
> www.bcgsc.ca
> grobertson at bcgsc.ca