User: Tim Peters
Tim Peters • 80
- Reputation:
- 80
- Status:
- Trusted
- Location:
- Australia
- Last seen:
- 2 months, 3 weeks ago
- Joined:
- 4 years, 3 months ago
- Email:
- t*******@garvan.org.au
--
Tim Peters, PhD
Bioinformatics Research Officer | Epigenetics Research Laboratory | Genomics and Epigenetics Division
Garvan Institute of Medical Research
384 Victoria St., Darlinghurst, NSW, Australia 2010
Posts by Tim Peters
<prev
• 37 results •
page 1 of 4 •
next >
0
votes
1
answer
1.6k
views
1
answers
... Hi Steven,
This is indeed possible using datatype="sequencing" - please look at the second half of the vignette for a worked example.
Best,
Tim
*Edit* My bad, I hadn't realised that the example included only contained output from DSS::DMLtest, not DSS::DMLtest.multiFactor. If you use output fro ...
written 6 months ago by
Tim Peters • 80
0
votes
0
answers
293
views
0
answers
... This is a problem with your version of zlib, not DMRcate. Looks like you need to update the version of zlib on your machine. This thread may be of help: https://www.biostars.org/p/89217/
Cheers,
Tim
...
written 12 months ago by
Tim Peters • 80
0
votes
0
answers
293
views
0
answers
... The extractRanges() function now appends promoter associations from Gencode/Ensembl instead. I find this annotation more reliable than Illumina's. Have a look at page 4 of the vignette:
https://bioconductor.org/packages/release/bioc/vignettes/DMRcate/inst/doc/DMRcate.pdf
Cheers,
Tim
...
written 12 months ago by
Tim Peters • 80
0
votes
2
answers
356
views
2
answers
Comment:
C: Bumphunting for longitudinal data
... Hi Ash,
Yes, it looks like that will work. The ellipsis argument (...) at the end of cpg.annotate() is matched to extra arguments for limma::lmFit(), so block and correlation will propagate to the linear fitting.
And yes, DMRcate can definitely be used for continuous responses as well. The only ca ...
written 14 months ago by
Tim Peters • 80
1
vote
2
answers
356
views
2
answers
... Hi Ash,
Yes, like James said, you'll need to sacrifice some of your covariates to get your design matrix full rank. Once you have a full rank matrix, you can then pass it to cpg.annotate() along with your coefficient of interest (and optionally contrast matrix), and then the regular workflow will d ...
written 14 months ago by
Tim Peters • 80
1
vote
3
answers
856
views
3
answers
... Hi Giovanni,
Any post-filtering decision is up to you - I've made DMRcate that way so user can have some control over what thresholding they believe is best. Many users I've encountered have made decisions in line with what you describe in your bullet points.
Best,
Tim
...
written 20 months ago by
Tim Peters • 80
0
votes
3
answers
856
views
3
answers
... Hi Poojitha,
Looks like the S4 structure of IlluminaHumanMethylationEPICanno.ilm10b2.hg19 has changed.
Try
locs <- IlluminaHumanMethylationEPICanno.ilm10b2.hg19::Locations
for the first line instead.
Cheers,
Tim
...
written 20 months ago by
Tim Peters • 80
1
vote
1
answer
534
views
1
answers
... Hi elleno,
Thanks for picking this up - we rewrote dmrcate() recently to speed it up a bit, but looks like we neglected to reorder the DMRs. I have put the fix in svn and it should propagate to the web within 48 hours, or you can grab the corrected version right now at https://github.com/timpeters8 ...
written 2.1 years ago by
Tim Peters • 80
0
votes
3
answers
1.0k
views
3
answers
... Hi Rahel,
Well, it says "Partial NA coefficients for 1 probe(s) " now instead of the 61,723 you had before, so that's progress. This means there is only one row in the matrix for which the model is inestimable.
I would check a few more things about mVals, you'll need to remove rows that have any ...
written 2.3 years ago by
Tim Peters • 80
0
votes
3
answers
1.0k
views
3
answers
... Come to think of it, it is vital that you name your column(s) in contMatrix, and then pass a column name to coef in your call to cpg.annotate(). Otherwise it will simply take the first column of design as your coefficient of interest, which I assume is not what you want. Look at what Jenny has done: ...
written 2.3 years ago by
Tim Peters • 80
Latest awards to Tim Peters
Scholar
2.1 years ago,
created an answer that has been accepted.
For A: Bumphunting for longitudinal data
Autobiographer
4.3 years ago,
has more than 80 characters in the information field of the user's profile.
Use of this site constitutes acceptance of our User
Agreement
and Privacy
Policy.
Powered by Biostar
version 16.09
Traffic: 174 users visited in the last hour