[BioC] error loading dba in DiffBind

[Gordon Brown]

Hi, Nathan,

I haven't seen messages like these.  The warnings suggest that numbers are
being interpreted as strings, and converted to factors, but I can't
imagine why that would happen.  Can you send along your sample sheet, and
the first few lines of your peaks files?  I'll see if I can reproduce it.
Also, can you let me know the sessionInfo() and operating system and
version from both your computer and cluster?

In case it helps, "dba.count" now has a "bLowMem" option that greatly
reduces memory usage in dba.count; perhaps you will be able to get further
on your local machine using that option.  You'll have to upgrade to R
3.0.1/Bioconductor 2.12 though.
 
Cheers,

 - Gord

On 2013-06-19 15:23, "Lawson, Nathan" <Nathan.Lawson at umassmed.edu> wrote:

>
>I am using DiffBind to identify differentially occupied elements from
>histone modification ChIP-Seq between two different cell lines.
>
>I successfully ran the package on my own computer with peaks and mapped
>reads limited to a single human chromosome as a test.  The analysis ran
>nicely and the results looked good.  Unfortunately, I was not able to run
>a full genome-wide analysis due to computational and space limitations.
>Therefore, I tried to run the analysis on our high performance computing
>cluster, which is when the error appeared.
>
>When running the SAME EXACT set of files, including the same sample
>sheet, on our cluster, I get the following error output, as well as
>additional warnings that I have not seen previously:
>
>> AV = dba(sampleSheet="AVbothChr6.csv")
>A1.0 HUAEC K27ac artery  1 raw
>A1.1 HUAEC K27ac artery  2 raw
>V1.0 HUVEC K27ac vein  1 raw
>V1.1 HUVEC K27ac vein  2 raw
>A1.2 HUAEC p300 artery  1 raw
>A1.3 HUAEC p300 artery  2 raw
>V1.3 HUVEC p300 vein  1 raw
>V2.6 HUVEC p300 vein  2 raw
>Error in if (res >= minval) { : missing value where TRUE/FALSE needed
>In addition: Warning messages:
>1: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>2: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>3: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>4: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>5: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>6: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>7: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>8: In Ops.factor(peaks[, pCol], width) : / not meaningful for factors
>> 
>
>
>Again, these datasets were successfully entered as a dba object and
>subsequently analyzed using DiffBind on my computer.  The original input
>files were simply transferred to the cluster to re-test there.  The only
>difference (I can see) is that the cluster is currently running R-3.0.1
>and I was running 2.15.
>
>I also tried making the peaks.bed files into a 6-column format (the
>original bed files only had 5 columns), but this did not seem to solve
>the problem.
>
>Any suggestions are welcome.
>
>Thanks,
>Nathan
>
>Nathan D. Lawson, Ph.D.
>Associate Professor
>Program in Gene Function and Expression
>University of Massachusetts Medical School
>364 Plantation Street
>LRB617
>Worcester, MA 01605
>website: lawsonlab.umassmed.edu
>email: nathan.lawson at umassmed.edu
>phone: 508-856-1177
>