-
Notifications
You must be signed in to change notification settings - Fork 22
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
How to implement more complex contrast formulas #305
Comments
Hi @CarlottaHS - are you looking for interactions between the |
Hi @MikeDMorgan, thanks for your quick reply! We want to study the effect of the |
Hi @CarlottaHS Sorry - I didn't get the notification for your response. If you have multiple samples from the same individual, then yes, these are dependent observations (they are dependent on the individual). Whether this has a big impact on your analysis isn't 100% clear, but it does mean they aren't independent - adjusting for the individual would help to alleviate this. It sounds like you need to do some exploratory data analysis, for example comparing V1 vs the grouped V2, V3 and V4. You can create a new dummy variable that has 1's for V1 and 0's otherwise. |
I am experiencing the same problems as well that gives me the identical results for by using Many thanks. |
Please don't cross-post issues - you have an issue open already in #333 |
Hi!
Thanks a lot for the great tool! We already got some interesting results, but now are running into issues using more complex design and contrast matrices.
I tried following your vignette in your “Making comparisons for differential abundance using contrasts” notebook and the limma syntax for specifying contrast and design matrix, but so far no success.
You point to the limma and edgeR bioconductor pages for in-depth discussion on deciding which design is useful. But now are running into problems when trying to use these designs.
We have a design matrix, which includes many different factors.
What works are simple contrasts:
and then running each contrast sequentially. E.g.:
But we further also want to include the timepoint/ Visit into the analysis and not just the diagnosis. Here, we run into issues. The code is running, but the results are not what we would expect.
We also tried using some other model matrices such as “~0 + Diagnosis + Diagnosis:Visit” etc. but the results are similar.
We are a bit lost as we are following the limma syntax and are using their tutorial to create our model and contrast matrices and they have the expected format, so we are wondering what we need to do differently for the Milo package.
I hope my explanations are clear and sufficient. If not I am happy to provide more context!
Thanks in advance!
The text was updated successfully, but these errors were encountered: