Moderator: Ryan C. Thompson

gravatar for Ryan C. Thompson
Reputation:
5,660
Status:
Trusted
Location:
The Scripps Research Institute, La Jolla, CA
Website:
http://stackoverflow.c...
Twitter:
@DarwinAwdWinne
Last seen:
2 days, 3 hours ago
Joined:
4 years, 6 months ago
Email:
r**@thompsonclan.org

Posts by Ryan C. Thompson

<prev • 666 results • page 1 of 67 • next >
0
votes
1
answers
40
views
1
answers
Comment: C: What is the proper way to modify mcols of a GRanges inside a GRangesList?
... The level="within" trick is exactly what I was looking for. I don't know if we really need a shortcut for this pattern, since it doesn't come up that often. As long as there's a way to do it at all without breaking abstraction, I'm satisfied. ...
written 2 days ago by Ryan C. Thompson5.7k
3
votes
1
answer
40
views
1
answer
What is the proper way to modify mcols of a GRanges inside a GRangesList?
... Sometimes I have a GRangeList object and I want to modify the mcols of the underlying GRanges object. The only way I know to do this is to do something like: grl@unlistData$newColumn <- newValue But that unfortunately breaks the GRangesList abstraction by directly accessing the slot. Is there ...
granges metadata written 2 days ago by Ryan C. Thompson5.7k • updated 2 days ago by Michael Lawrence9.2k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... You never specified that the null hypothesis was true for the gene in your thought experiment. You simply said that you could take any arbitrary gene X, repeat the experiment 1000 times, and use the results to derive an empirical null distribution with no mention of the true differential expression ...
written 4 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... Thousands of p-values from thousands of repetitions of the same experiment will not converge to any value. Nor will their distribution converge to a uniform distribution unless the null hypothesis for that experiment is true. ...
written 5 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... p-values are generated by comparison to a theoretical null distribution (or sometimes an empirical null via permutation tests, etc.). You can't, as you suggest, treat the test statistics from 100 repetitions of an experiment as a null distribution from which to calculate p-values for those test stat ...
written 5 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... Refer to Mike Love's reply for more reading, but the short answer is that information about the variance of the rest of the genes in the genome can be used to get a more stable estimate of the variance for gene X than you could obtain using only the data for gene X. Remember that you never know the ...
written 5 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
83
views
2
answers
Answer: A: limma for Paired Samples vs. lme
... I'm not familiar with the model specification syntax of lme, but it looks like you're treating Donor as a random effect in lme and a fixed effect in limma. limma can fit a mixed effects model with one random effect term using the duplicateCorrelation function. I would expect this to give you p-value ...
written 6 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
102
views
2
answers
Comment: C: Alternative packages to limma for analysis of microarray datasets
... By my understanding, the alternatives to limma for RNA-seq, such as edgeR and DESeq, were created because limma's statistical model was not (at that time) a good fit for count data. Do you have concerns about limma's statistical model for your data, or is there a specific statistical method you're l ...
written 6 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... I think it's more than a bit disingenuous to declare that you have major concerns and request that someone else to spend time and effort to validate those concerns while at the same time declaring that same validation work to be too much trouble for you, especially since in this case both the data a ...
written 7 days ago by Ryan C. Thompson5.7k
0
votes
2
answers
379
views
2
answers
Comment: C: Are published RNA seq data analyses often wrong in calculating p-values and FDR?
... Everyone is well aware that the models being used are a simplification of the truth. This is the nature of a statistical model, hence the classic aphorism "All models are wrong but some are useful". You're saying that the negative binomial assumption is questionable, but then you say you prefer a t- ...
written 7 days ago by Ryan C. Thompson5.7k

Latest awards to Ryan C. Thompson

Teacher 18 days ago, created an answer with at least 3 up-votes. For A: What's the relationship between "design" and "contrasts"?
Scholar 18 days ago, created an answer that has been accepted. For A: What's the relationship between "design" and "contrasts"?
Teacher 10 weeks ago, created an answer with at least 3 up-votes. For A: What's the relationship between "design" and "contrasts"?
Scholar 10 weeks ago, created an answer that has been accepted. For A: What's the relationship between "design" and "contrasts"?
Scholar 11 weeks ago, created an answer that has been accepted. For A: What's the relationship between "design" and "contrasts"?
Teacher 11 weeks ago, created an answer with at least 3 up-votes. For A: What's the relationship between "design" and "contrasts"?
Scholar 12 weeks ago, created an answer that has been accepted. For A: What's the relationship between "design" and "contrasts"?
Scholar 12 weeks ago, created an answer that has been accepted. For A: What's the relationship between "design" and "contrasts"?
Teacher 3 months ago, created an answer with at least 3 up-votes. For A: What's the relationship between "design" and "contrasts"?
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: DESeq2 how to define my conditions when I want the results comparing a subset of
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: Usability of ERCC spike-ins in standard RNAseq experiments
Scholar 13 months ago, created an answer that has been accepted. For A: normalized data by DESeq2 look suspicious
Scholar 13 months ago, created an answer that has been accepted. For A: On the relationship between design and contrasts
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: Accounting for 5'/3' Bias in DESeq 2
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: FDR and padj in deseq and edger
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: edger, trended or common dispersion
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: edgeR RNA-seq analysis using glm
Good Answer 13 months ago, created an answer that was upvoted at least 5 times. For A: Using multiple tools for differential gene expression analysis as validation
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: DE for genes with very low counts using limma.
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: Ribosome profiling analysis in DEseq2/limma
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: Using multiple tools for differential gene expression analysis as validation
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: On the relationship between design and contrasts
Teacher 13 months ago, created an answer with at least 3 up-votes. For A: What's the relationship between "design" and "contrasts"?

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 144 users visited in the last hour