Question: AnnotationHub: confusing pValue/qValue for (some) Roadmap records
1
3.6 years ago by
kevin.rue230
University of Oxford
kevin.rue230 wrote:

Dear all,

library(AnnotationHub)
ah = AnnotationHub()
X = c("AH29560","AH29561","AH29562","AH29563","AH29564","AH29565"),
FUN = function(record){ah[[record]]})


To my surprise, the qValue column shows larger -log10 values than the pValue column, e.g.:

GRanges object with 654611 ranges and 6 metadata columns:
seqnames               ranges strand |        name     score signalValue    pValue    qValue      peak
<Rle>            <IRanges>  <Rle> | <character> <numeric>   <numeric> <numeric> <numeric> <numeric>
[1]    chr16 [58585453, 58585749]      * |      Rank_1        90        <NA>   6.75724  11.99359   9.03636
[2]    chr18 [12985202, 12985439]      * |      Rank_2        88        <NA>   7.13022  11.78810   8.83747

I have looked at http://genome.ucsc.edu/FAQ/FAQformat.html#format13 to try and understand, but I remain confused.

I will also contact directly the Roadmap project for this, but I thought that 1) someone here might be able to clarify this for me, and 2) it may be of interest to other people here.

Kevin

modified 3.6 years ago by Valerie Obenchain6.7k • written 3.6 years ago by kevin.rue230
2
3.6 years ago by
United States
Valerie Obenchain6.7k wrote:

Hi,

Thanks for reporting the bug. These files were stamped with a dispatch class that wasn't distinguishing between narrow and broad peak files which have different metadata columns. The narrow peak files have an extra 'peak' column as shown in the ucsc link you reference. Because of this the metadata columns of the broad peak files were shifted to the right - what was really 'qvalue' was called 'peak', what was really 'pvalue' was called 'qvalue'.

A fix is checked into AnnotationHub 2.3.20 and should be available Saturday after noon PST. Let me know if you have other problems.

Valerie