## User: Axel Klenk

Axel Klenk920
Reputation:
920
Status:
Trusted
Location:
Switzerland
Last seen:
6 hours ago
Joined:
9 years, 6 months ago
Email:
a*********@idorsia.com

#### Posts by Axel Klenk

<prev • 121 results • page 1 of 13 • next >
1
51
views
1
... I can reproduce this behaviour when I have pkg rlang attached and can see it in your sessionInfo() as well. rlang masks exprs(). The easiest way to get around this problem is to use > str(Biobase::exprs(gse[[1]])) Does that work for you?   ...
written 15 hours ago by Axel Klenk920
1
51
views
1
... See the Value: section of ?getGEO : it returns a list of ExpressionSet objects, not an ExpressionSet object. Hence, you want: > str(exprs(gse[[1]])) Hope this helps.   ...
written 3 days ago by Axel Klenk920
1
45
views
1
... (Hmmm... not sure what happened to my previous answer. Anyway:) No, DESeq2 is not normally loaded on R startup. I think this could be an issue with your RStudio project or R may restore a previously saved worksapce on startup that contains a DESeq2 object. The latter can be avoided by starting R li ...
written 5 days ago by Axel Klenk920
1
43
views
1
... Apparently the installation failed. What is the output from your biocLite("affydata") ?   ...
written 5 days ago by Axel Klenk920
1
134
views
1
... Good. 1) I cannot tell only form the PCA plot. 2) Have a look at Bioc pkg sva. Try finding surrogate variables in your data (see the vignette). If you find one that you're confident is a batch effect, you can try removing it using sva or combat (same pkg). Or, in case you want to use limma for dif ...
written 6 days ago by Axel Klenk920
1
134
views
1
... Yes, as a consequence of independent filtering some padj are NA, so you need > write.csv(as.data.frame(res[!is.na(res$padj) & res$padj < 0.1 & res$log2FoldChange > log2(1.5),]), file="DEGs.csv") Hope this will work. :-) ... written 6 days ago by Axel Klenk920 1 answers 134 views 1 answers ... Well, consider the pros and cons: if your screening step is too conservative, you may not have enough candidates for your wet lab experiments and you have a higher risk of missing false negatives. If, OTOH, it is not conservative at all, you may end up with too many candidates for wet lab follow-up ... written 6 days ago by Axel Klenk920 1 answers 134 views 1 answers ... > results(dds, lfcThreshold = log2(1.5), alpha = 0.1) does not do what you apparently think it does, see ?results . You probably want something like > write.csv(as.data.frame(res[res$padj < 0.1 & res\$log2FoldChange > log2(1.5),]), file="DEGs.csv") ...
written 6 days ago by Axel Klenk920
1
134
views
1
... Yes, if I read the figure legend and paper correctly, they did this as a first step to select candidates that were then tested in additional experiments, and the thresholds were apparently chosen to yield a manageable number of candidates. Why not. So, let me clarify, that I would not recommend usin ...
written 6 days ago by Axel Klenk920
1
134
views
1
... 1) Yes, in which case the resulting list of "DEGs" is expected to contain 10% false discoveries instead of 5%. 2) Sure, but you may want to check that the lfc of your "DEGs" is biologically meaningful in the context of your experiment and not "only" statistically significant. 3) I would not recomm ...
written 6 days ago by Axel Klenk920

#### Latest awards to Axel Klenk

Teacher 12 months ago, created an answer with at least 3 up-votes. For A: Limma un moderated t-test
Centurion 12 months ago, created 100 posts.
Scholar 12 months ago, created an answer that has been accepted. For A: Identification of DEGs through limma analysis
Scholar 12 months ago, created an answer that has been accepted. For A: Limma un moderated t-test
Scholar 2.3 years ago, created an answer that has been accepted. For A: Limma un moderated t-test
Supporter 2.4 years ago, voted at least 25 times.
Teacher 3.5 years ago, created an answer with at least 3 up-votes. For A: basic R question: concatenate two numeric vectors grouped by element index

Content
Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.