Vignette

[ChristinaSchmidt1]

1. Use The nature comms CoRe data and check if there is intra and media.
2. Implement the data used for the vignette as a callable R dataframe
3. Write the vignette

Otherwise: Use CCLE data.

[ChristinaSchmidt1]

• Add an example of the pre-selection of metabolites to be plotted on e.g. the lollipopgraph. This is needed if there is many metabolites as this will over crowd the plot.

• Add tables and overviewalluvial plot explaining the general concept of the metabolite clusters to the vignette or R documentation.

[ChristinaSchmidt1]

Disscuss 80% filtering rule and suggest user to change default in specific cases:

Removing samples were more than 20% have no detection is way to stringend when thinking about biology.

Modified 80% filtering rule: If you do it group-wise, which for me would mean on the biological background, e.g. FH-KO and FH-WT, it is quite stringent. If we have 5 replicates and one is NA for a metabolite according to this rule we would remove it. Yet I think thats a bit too stringent and I would set the default to 40%, meaning 1-2 out of 5 samples could have NA.

Standard 80% filtering rule: Done on all samples, hence we assume that there is no groups, so here we can be more stringent.

But thinking further, if one has patients data, where there can be many unknown subgroups within tumour samples (e.g. due to gender, age, stage,...), which leads to a metabolite being detected in only 50% (or even less) of the tumour samples, yet with the rule we would remove this. So here we probably do want to set a default, but also give the user a message if there is many metabolites removed.

[ChristinaSchmidt1]

I have done all of this and the remaining isues are in regards to making website with pkgdown, so there are listed in an new issue