User: gabriel.hoffman

Reputation:
80
Status:
Trusted
Location:
United States
Website:
http://gabrielhoffman....
Scholar ID:
Google Scholar Page
Last seen:
1 week, 2 days ago
Joined:
4 years, 3 months ago
Email:
g**************@mssm.edu

I develop and apply statistical methods in genomics (mostly RNA-Seq) to understand the biology of human disease.

Posts by gabriel.hoffman

<prev • 14 results • page 1 of 2 • next >
1
vote
3
answers
101
views
3
answers
Answer: A: dream vignette says limma-voom repeated measures two-group comparison needs two
... When I referred to two rounds of `duplicateCorrelation()`, I was referring to Gordon Smyth's suggestion rather than the original limma manual. The discrepancy is not ideal. Since `voom()` uses `lmFit()` internally to get the residuals, specifying a `block` will be used to compute the residuals and ...
written 9 days ago by gabriel.hoffman80
0
votes
3
answers
133
views
3
answers
Answer: A: Results from variancePartition not matching those from lme4
... Hi c.taylor, Thanks for your bug report. I have fixed the issue in the latest release 1.14.1. Let me know if you find any other issues -- Gabriel ...
written 15 days ago by gabriel.hoffman80
0
votes
2
answers
139
views
2
answers
Answer: A: Blocking factor in limma/voom vs. variancePartition/dream analysis.
... Hi Ben, I wanted to get back to you about your earlier questions. The new version 1.14.1 implements voomWithDreamWeights() so you can specify random effects in the calculation of the precision weights - Gabriel ...
written 15 days ago by gabriel.hoffman80
2
votes
3
answers
133
views
3
answers
Answer: A: Results from variancePartition not matching those from lme4
... That is certainly not ideal. The strange result is due to the fact that linear mixed model are solved with iterative algorithms that depend on: 1. the starting point of the parameters values and 2. the convergence criteria variancePartition fits the model on the first gene and then uses this ...
written 6 weeks ago by gabriel.hoffman80
1
vote
3
answers
133
views
3
answers
Answer: A: Results from variancePartition not matching those from lme4
... Hi c.taylor, Thanks for your question. It turns out that the one difference is the estimated value of the residual variance. First I will talk about estimiating this terms on linear models before extending to linear mixed models. # Start with a simple linear model # set seed for repr ...
written 7 weeks ago by gabriel.hoffman80
0
votes
3
answers
133
views
3
answers
Comment: C: Results from variancePartition not matching those from lme4
... I will test this out and get back to you this week. I have fixed a few bugs in my development version, so its possible that this resolved in the latest version I'm planning to post next week. Cheers, Gabriel ...
written 7 weeks ago by gabriel.hoffman80
0
votes
1
answer
151
views
1
answers
Comment: C: BiocParallel: What its the best way to pass large a dataset for parallel process
... I really appreciate your detailed answer Cheers, - Gabriel ...
written 12 weeks ago by gabriel.hoffman80
0
votes
1
answer
151
views
1
answers
Comment: C: BiocParallel: What its the best way to pass large a dataset for parallel process
... Hi Martin, Thanks for your quick response! Yes, this was just example code, so `lmFit()` would solve this specific problem, but not my real problem. Your last paragraph gets at my underlying question: How do I get `bplapply()` to run in parallel when I pass it an iterator? Can I define my itera ...
written 12 weeks ago by gabriel.hoffman80
1
vote
1
answer
151
views
1
answer
BiocParallel: What its the best way to pass large a dataset for parallel processing?
... I have a large gene expression matrix and would like to perform some analysis on each row (i.e. gene) at a time using parallel processing from `BiocParallel`. Since the matrix can be very large, I figured that using `bplapply()` with `SnowParam(workers=12, "SOCK")` and a custom iterator would enabl ...
biocparallel bioconductor bplapply bpiterate iterator written 12 weeks ago by gabriel.hoffman80 • updated 12 weeks ago by Martin Morgan ♦♦ 23k
2
votes
2
answers
139
views
2
answers
Answer: A: Blocking factor in limma/voom vs. variancePartition/dream analysis.
... Hi Ben, Thanks for your comment. Both limma and dream analysis include two steps: 1) estimate weights, 2) fit regression model using these weights. The limma/voom workflow uses `voom()` and then `lmFit()`. The dream package focuses on replacing `lmFit()` with `dream()` in order to model random e ...
written 3 months ago by gabriel.hoffman80

Latest awards to gabriel.hoffman

Scholar 10 months ago, created an answer that has been accepted. For A: Blocking factor in limma/voom vs. variancePartition/dream analysis.
Autobiographer 3.3 years ago, has more than 80 characters in the information field of the user's profile.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 341 users visited in the last hour