I am in the process of study design and wanted to inquire if the following would be viable for use in this platform or if there was another direction I should look. The current design is 3 samples for two genotypes at 8 time points across the day. It is intended to use bulk RNA-seq, proteomics and metabolomics to be performed on each sample. Is this viable to integrate the three datasets using this platform? The aim is to identify candidate mechanisms as canonical or non-canonical and evaluate them
Yes mixOmics is able to integrate transcriptomics, proteomics and metabolomics datasets together. I am assuming from your experimental design you may be interested in comparing your two genotypes, so you could use a supervised model such as DIABLO. Another tool you might be interested in is TimeOmics which you can use to identify correlated profiles over time and between omics.
Thank you for your recommendations. Will these be viable for our use with low biological repetitions, I think we have increased it to 4 per genotype per time point. The comparisons are genotype, time and genotypeXtime as it is a circadian study. I will look into your suggestions but it seems DIABLO is for separate pools of samples per omic rather than the same sample split across all omics (is this P-integration or am I confused?). Do we have enough counts for supervised to function properly? Apologies my baseline of knowledge with networks is a little playing around in WEKA with their demo databases and it seems their counts are much much higher.
I can only comment generally on sample size as it will depend on your data and your experimental aim, if you see a very clear pattern in your omics data 4 samples per condition might be sufficient, but if the data is noisy (as unfortunately is the case with biological data most of the time!) it will be harder to find lots of variables, and therefore mechanisms, that differ between your genotypes and timepoints. I would therefore usually recommend increasing sample size as much as possible (10 or so would be good), but this is just a general recommendation and may not be appropriate for your particular study.
DIABLO works for exactly that - where the same samples have been used to generate multiple omics, so I think this would work well for you. I would highly recommend checking out our mixOmics website for more information to help you design your experiment and analysis plan, this page in particular gives an overview of the different mixOmics methods and what they can be used for.