User: Bernd Klaus

gravatar for Bernd Klaus
Bernd Klaus540
Reputation:
540
Status:
Trusted
Location:
Germany
Twitter:
@BKlausStats
Last seen:
6 months ago
Joined:
5 years, 4 months ago
Email:
b**********@embl.de

Posts by Bernd Klaus

<prev • 63 results • page 1 of 7 • next >
3
votes
2
answers
275
views
2
answers
Answer: A: Dispersion fit and p-value distributions
... Hello, thanks a lot for your very detailed post! As you noted yourself, the parametric fit seems to slightly overestimate the dispersions for your data, so the p-values you get   from the Wald test z-scores are too large on average, fdrtool re-estimates the variance of the z-scores and gives you a ...
written 6 months ago by Bernd Klaus540
0
votes
2
answers
1.1k
views
2
answers
Comment: C: DESeq2 dispersion and p-value histograms
... You can also try to compare the dispersion estimates for the different time points directly as the prior comparison assumes that the trend fit is similar (I edited my answer accordingly). They can be accessed via rowData(dds)$dispersion. And you could e.g. look at a scatterplot of log10(rowData(dds) ...
written 17 months ago by Bernd Klaus540
2
votes
2
answers
1.1k
views
2
answers
Answer: A: DESeq2 dispersion and p-value histograms
... Hi Ronald, as Mike said, there is always some "detective work" involved on determining why a p-value distribution doesn't look like it "should".    In my workflow, I validated the correction by looking at the overlap of the results of the original analysis with my corrected list, where I find 105 ...
written 18 months ago by Bernd Klaus540
0
votes
1
answer
413
views
1
answers
Comment: C: Error when restore rownames of Feature data of ExpressionSet
... I agree with Gavins's answer - the code will exclude all probe IDs with multiple mappings to gene symbols, so by definition there should not be any duplicate PROBEIDs in fData as I retain only the PROBEIDs matching to exactly one gene symbol. Are you using the correct annotation database for your a ...
written 18 months ago by Bernd Klaus540
0
votes
1
answer
866
views
1
answers
Comment: C: function getAE from ArrayExpress needs update because EBI has switched to https
... The team will update the function. In the meantime you can simply use my function https_getAE in the the github gist linked to above in order to make it work again. ...
written 18 months ago by Bernd Klaus540
3
votes
1
answer
866
views
1
answer
function getAE from ArrayExpress needs update because EBI has switched to https
... Dear ArrayExpress development team, EBI services (<https://www.ebi.ac.uk/services>), including ArrayExpress have switched to https on October 1st 2017.  The function getAE from the ArrayExpress Bioconductor package needs an update, as any retrieval fails like so: library(ArrayExpress) getA ...
arrayexpress rcurl xml http ebi written 18 months ago by Bernd Klaus540 • updated 18 months ago by Wolfgang Huber13k
1
vote
1
answer
1.1k
views
1
answers
Answer: A: DESEq2 p-values weird distribution under the H0 and fdrtool
... Hi Thyago,   The PCA plot looks pretty much like a batch effect being a stronger "separator" than the experimental groups. You should tackle this directly. After application of fdrtool, you move to a rather bathtub shaped p-value histogram which often indicates p-values that are too liberal (you ...
written 20 months ago by Bernd Klaus540
0
votes
2
answers
961
views
2
answers
Answer: A: Alternative packages to limma for analysis of microarray datasets
... if you are interested in two group comparisons only, the st CRAN package implements various alternative variance shrinkage estimators, the CAT score even takes correlations between the genes into account:   http://strimmerlab.org/software/st/index.html   ...
written 23 months ago by Bernd Klaus540
0
votes
2
answers
1.3k
views
2
answers
Answer: A: Inspection and Correction of p-values in DESeq2 pipeline by Bernd
... If you are sure that there are no batch effects, try to run a stronger prefiltering of your data, e.g. rowSums(counts(dds)) > 20 or something like that since you had spikes near 1 in the p-value histograms you sent to my by mail. The fdrtool results cannot directly indicate whether there is ...
written 2.1 years ago by Bernd Klaus540
0
votes
2
answers
1.3k
views
2
answers
Comment: C: Inspection and Correction of p-values in DESeq2 pipeline by Bernd
... As Mike said, this is a quite subjective decision. If the p-value is hill shaped, the p-values are on avarage "higher than expected" leading to a potentially to a lower power. However, Wolfgang pointed me to the fact that these histogram shapes can result from batch effects in your data that overla ...
written 2.1 years ago by Bernd Klaus540

Latest awards to Bernd Klaus

Popular Question 14 months ago, created a question with more than 1,000 views. For oligo annotation: pd.mogene.2.0.st vs pd.mogene.2.0.st.v1
Teacher 14 months ago, created an answer with at least 3 up-votes. For A: Problem using sva with limma for microarray DE analysis
Scholar 3.0 years ago, created an answer that has been accepted. For A: Problem using sva with limma for microarray DE analysis
Scholar 3.0 years ago, created an answer that has been accepted. For A: Help with DESeq2 Multiple Factor Design
Scholar 3.0 years ago, created an answer that has been accepted. For A: Trying to account for gene length in topGO analysis
Teacher 3.0 years ago, created an answer with at least 3 up-votes. For A: Problem using sva with limma for microarray DE analysis
Scholar 4.1 years ago, created an answer that has been accepted. For A: Help with DESeq2 Multiple Factor Design

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 185 users visited in the last hour