N-integration with 10 datasets

Hi there,

Thanks for developing mixOmics. It is a really great tool and well documented.
I intended to perform an N-integration analysis of 10 post-translational modification(PTM) datasets using mixOmics. My queries are as follow:

  1. To integrate 10 datasets(such as phosphorylation, acetylation, ubiquitination, succinylation, et al.), do you have any suggestions on how to perform the Tuning keepX step? In my case, the exponentially increasing number of variables resulted in too much calculation, and the Tuning process could not be completed.

  2. When I performed circosPlot visualization, there seems to be a limit of no more than 6 colors for block display. Does it mean that I cannot directly export visualization results using mixOmics?

Hi @Lou,

  1. Did you get an error? And did you try with multiple cpus (cpus argument)? My experience is that the tune.block.splsda is very good at using multiple cpus. In fact, i use up to 80 cpus because it runs much smoother/faster.

  2. Iā€™m not sure if this will work, but you might be able to add more colors by changing the color.blocks arguments.Try with color.blocks = color.mixo(1:10) (i think 10 colors is the maximum).

Let me know if this solved the issues.

1 Like

Thanks for your reply and suggestion. I will try with multiple cpus soon.
I tried visualization with this code(color.blocks = color.mixo(1:10)), but only 6 blocks are colored.

Hi @Lou,

Thank you for letting us know about this serious issue. It is now fixed in the latest GitHub versions which you can install using instructions here: https://github.com/mixOmicsTeam/mixOmics#github-versions

You can explore an example here: https://github.com/mixOmicsTeam/mixOmics/issues/96