User: Peter Hickey
Peter Hickey • 230
- Reputation:
- 230
- Status:
- Trusted
- Location:
- Australia
- Website:
- http://www.peterhickey...
- Twitter:
- PeteHaitch
- Scholar ID:
- Google Scholar Page
- Last seen:
- 6 days, 2 hours ago
- Joined:
- 4 years, 8 months ago
- Email:
- p***********@gmail.com
Posts by Peter Hickey
<prev
• 54 results •
page 1 of 6 •
next >
0
votes
3
answers
174
views
3
answers
... There were multiple warnings following bsseq.data.smoothed <- BSmooth(bsseq.data, verbose=TRUE). Did you look to see what there were? Above I suggested that because you are analysing RRBS data you would need to explore different parameters to BSmooth() but you've used the defaults. I suspect this ...
written 6 days ago by
Peter Hickey • 230
0
votes
1
answers
74
views
1
answers
... Hi Martin,
combineList(hdf5_BS1, hdf5_BS2, BACKEND = "HDF5Array") will give you a HDF5Array-backed BSseq object. It's somewhat buried in the docs, but see ?combineList.
Currently, the default location for HDF5 files is specified via HDF5Array::setHDF5DumpDir(). You can check the current dump dir ...
written 21 days ago by
Peter Hickey • 230
0
votes
3
answers
174
views
3
answers
... I am unable to reproduce the error. However, your example code has a bug: list.files(pattern="*.cov") -> cov.files means that the chr16 and chr20 data will be considered to have come from separate samples. This is not what you want.
I tidied up and simplified your example code and was able to su ...
written 4 weeks ago by
Peter Hickey • 230
0
votes
3
answers
174
views
3
answers
... Thanks, Nastasia. Can you please also post the exact command you ran that worked for chr20 but not for chr16
...
written 5 weeks ago by
Peter Hickey • 230
0
votes
3
answers
174
views
3
answers
... It's difficult to know what's causing the error from that output. Are you able to share the chr16 data with me (e.g., via Dropbox)? I could then investigate further
...
written 5 weeks ago by
Peter Hickey • 230
0
votes
3
answers
174
views
3
answers
... 1. Setting ns so large is probably not a good idea for RRBS data. As documented, ns is the minimum number of methylation loci in a smooth window. Most RRBS captured regions will contain many, many fewer CpGs, so you will be smoothing over large windows containing several clusters of CpGs with large ...
written 5 weeks ago by
Peter Hickey • 230
0
votes
3
answers
174
views
3
answers
... To answer some of your specific questions:
1. The warnings generated during BSmooth() are coming from the underlying locfit R package. Such warnings are quite technical by nature and can be hard to understand, but they often indicate a problem with the choice of smoothing parameters. The default BS ...
written 6 weeks ago by
Peter Hickey • 230
0
votes
1
answers
158
views
1
answers
... I'm currently exploring parameter choices for non-CG loci, but in human. I recognise that non-CG loci are prevalent and important for methylation in plants and that it would be good to provide recommendations. The simple reason we don't is because neither Kasper (main developer) nor myself analyse p ...
written 11 weeks ago by
Peter Hickey • 230
0
votes
1
answers
158
views
1
answers
... Hi Brandon,
This issue is that the default BSmooth() parameters, especially ns, h, and maxgap, are tuned for the distribution of CG loci in human/mammalian genomes [1]. You are using these default parameters to analyse CHH loci from what I'm guessing is soybean (based on 'Gm01' in your example), an ...
written 11 weeks ago by
Peter Hickey • 230
0
votes
1
answers
158
views
1
answers
... Thanks, Brandon. I was able to reproduce the error using the BS1.12M.CHH.gz file. Ultimately, an error is occurring when locfdr::locfit() is called from within bsseq::BSmooth(). Unfortunately, because it's a little buried and calling some fairly complicated code in locfdr, this means it will take a ...
written 3 months ago by
Peter Hickey • 230
Latest awards to Peter Hickey
Scholar
18 days ago,
created an answer that has been accepted.
For A: t-statistic cutoff for finding DMRs with bsseq
Scholar
16 months ago,
created an answer that has been accepted.
For A: t-statistic cutoff for finding DMRs with bsseq
Scholar
16 months ago,
created an answer that has been accepted.
For A: t-statistic cutoff for finding DMRs with bsseq
Use of this site constitutes acceptance of our User
Agreement
and Privacy
Policy.
Powered by Biostar
version 2.2.0
Traffic: 270 users visited in the last hour