Adding some multivariate approaches on MetaboAnalyst

MetaboAnalyst
, ,
1

Dear MetaboAnalyst users and software developers,

I would like to ask if you are planning to add to MetaboAnalyst the option to carry out statistical analysis on two factors. In omics are, some approaches (such as Anova Multiblock OPLS) are really useful to evaluate the discriminant biomarker compounds according to the potential interaction of two main factors included in the experinental plan. The final outcome provides the performance of the prediction model and the VIP^2 of each metabolite.

Addittionally, another very useful approach could be represented by the possibility to run OPLS-DA prediction models not onyl on 1vs1 groups, but also considering more than 2 groups. It could be nice to carry out on MetaboAnalyst unsupervised, supervised and Volcano Plot analysis on all the possible comparisons within the dataset (not only 1vs1), i.e. considering simoultaneously all the possible comparisons.

I hope this could be feasible in the next months.

Kind regards,
Gabriele

2

We are aware of many options when it comes to statistical analysis. The two main limiting factors are:

  • The computing resources - due to large user traffic, the public server (community version) is mainly for real-time computing (except raw spectra processing);
  • Our developers’ time - we are mainly funded for research-related development.

If you really want this feature, a more sensible way is to raise this request as a VIP (subscribed) user, as the feature does not fit the public server. We offer pro option which allows subscribed researchers to create and run a workflow → generate reports and slides on separate cloud nodes. You can see more details here.