User: Gavin Kelly

gravatar for Gavin Kelly
Gavin Kelly310
Reputation:
310
Status:
Trusted
Location:
United Kingdom / London / Francis Crick Institute
Twitter:
gavinpaulkelly
Scholar ID:
Google Scholar Page
Last seen:
3 hours ago
Joined:
2 years, 11 months ago
Email:
g*************@gmail.com

Statistician in the Bioinformatics & Biostatistics group at the Francis Crick institute

Posts by Gavin Kelly

<prev • 69 results • page 1 of 7 • next >
0
votes
0
answers
32
views
0
answers
Comment: C: Differential expression of nomalised and log2 transformed data in limma
... It looks like you're using the wrong function.  Limma provides 'lmFit', not the lm.fit function you're using, which is something entirely different.  Also, it's probably worth turning Array_tumorandstroma2 into a numeric matrix, rather than the data.frame that read.table returns.  And for safety, I' ...
written 1 day ago by Gavin Kelly310
0
votes
2
answers
82
views
2
answers
Comment: C: About the results and contrast (DESeq2)
... Having the normal samples in the deseq object, even if you're not using them explicitly in the contrast, adds a few extra degrees of freedom better to estimate the variance of the residuals.  So it then comes down to how much you believe the within-normals variability is consistent with the within-t ...
written 1 day ago by Gavin Kelly310
0
votes
2
answers
82
views
2
answers
Answer: A: About the results and contrast (DESeq2)
... For the "differentially expressed in metastasis vs primary tumor" you can get that directly by  results(ds.deseq, contrast=list(c("sample_typemetastasis"), c("sample_typeprimary"))) (approach A) because the sample_typemetastasis, and sample_typeprimary are relative to normal (that being the ba ...
written 4 days ago by Gavin Kelly310
0
votes
1
answers
120
views
1
answers
Answer: A: DESeq2: Is it necessary to include all terms and interactions in LRT tests?
... It looks to me like you've got the correct model specification, your reduced model should include all the nuisance terms that you want to normalise out of your data; the full model should include those, plus the (usually one) term (either main or interaction), that you hypothesize might be having an ...
written 9 days ago by Gavin Kelly310
0
votes
1
answers
112
views
1
answers
Answer: A: deseq2 LRT vs Wald test results for DE expression among 4 conditions
... I think this is due to the nature of the difference between LRT and Wald tests.  LRT is testing against all conditions being the same, so if the conditions are all slightly different, this could be enough for the LRT to reject the null, but none of the separate pairwise comparisons could reach signi ...
written 10 days ago by Gavin Kelly310
0
votes
1
answers
102
views
1
answers
Answer: A: DESeq2 with 4 groups, 5 timepoints and more...!
... This might be worth getting a local statistician on board for, because there are multiple subtly different approaches depending on precisely what you mean.  For example, in question 1, let's just take the case of CSF, so we subset the data down to 24hr & CSF.  We could then take a model ~ severi ...
written 10 days ago by Gavin Kelly310
0
votes
1
answers
390
views
1
answers
Answer: A: DESeq2 multifactorialdesign contrasts
... There are a couple of reasons I can think of that may explain this, but I assume your 75-gene list is fairly similar in content to the 85-gene list (if not, then the most likely explanation is that the data-importing has got sample labels misaligned).  The most likely cause is due to the way replic ...
written 6 weeks ago by Gavin Kelly310
0
votes
1
answers
395
views
1
answers
Answer: A: Difference between DESeq2 1.14 and 1.16
... You can use dds <- DESeq(dds, betaPrior=TRUE) to get the resultsNames to match, but I think if you're doing an LRT test, then it shouldn't make much difference, as the full and reduced models are specified in exactly the same way as before, without referencing specific coefficients.  And if yo ...
written 6 weeks ago by Gavin Kelly310
0
votes
2
answers
377
views
2
answers
Answer: A: deseq2 complex time course data with out replicates
... While completely agreeing with Michael's answer on the impossibility of testing pairs of timepoints, I'm wondering if you could take some prior expectations into account.  If you're expecting WT and KO to behave similarly at certain timepoints (may be the 10min placebo), then you could label those s ...
written 8 weeks ago by Gavin Kelly310
0
votes
2
answers
126
views
2
answers
Answer: A: [DESeq2] time-course Experimental design
... Still a bit confused about your use of LRT in the 'cheaper' design - it does not look at 'all... timepoints versus the 0h' - it's looking to see if the null of all timepoints being the same holds.  So an LRT could feasibly come out as significant if the end timepoint was different from all previous ...
written 8 weeks ago by Gavin Kelly310

Latest awards to Gavin Kelly

Scholar 6 months ago, created an answer that has been accepted. For A: DESeq2 3-fatcor design and different interaction terms
Teacher 6 months ago, created an answer with at least 3 up-votes. For A: extracting the genes associated with the clusters
Autobiographer 6 months ago, has more than 80 characters in the information field of the user's profile.
Scholar 22 months ago, created an answer that has been accepted. For A: DESeq2 3-fatcor design and different interaction terms

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 134 users visited in the last hour