Highlighting reviews - documentation and readme links

[mallewellyn]

Hi!

As Ailith and I have discussed, I've collated all the feedback from reviews and instructors & The Carpentries guidelines so that it's clear what's left to be done both to us and Carpentries reviewers.

Hopefully this is helpful but very happy to change the format or text, whatever's useful! Also, I essentially just went through the entire repo and picked out all reviews and feedback. I could well have missed some within the repo or any stored externally, so happy to edit if you can see some are missing.

I've also edited the readme slightly to link to the reviews, but also changed some of the initial text as I think this was just a default from the initial clone? Of course, happy to revert.

Hopefully seeing the list of to dos isn't too annoying - I'll try make some good progress with these!

[mallewellyn]

Have tried to create a data page as far as I can with my primitive biosciences knowledge. Will absolutely need checking and filling out in places!!

[alanocallaghan]

Looks good to me but will leave the final word for @ailithewing.

I think re:

• The lesson does not make use of superfluous data sets: ensuring that later uses of microarray and scrnaseq data is justified (as opposed to prostate/methylation).

that ideally one could probably just use the methylation data for everything without much loss, as long as the results aren't totally contrived. This might reduce the cognitive load a bit from hopping dataset to dataset every couple of hours. However it's a bit of a stretch goal right now to say the least as it would require a hell of a lot of rewriting

[mallewellyn]

Agreed, I think using the methylation data would be good. However, on the whole, I don't think the current data sets are 'superfluous' in that it's probably also good to strike a balance between seeing how these methods behave on different examples and cognitive load. There are some small examples where another data set is introduced where it probably doesn't need to be though, e.g. the use of cars, which also feels a bit random from a biosciences perspective. I'll take another look to see if I can propose minor changes that would sort this.

[mallewellyn]

Also, thank you for blitzing the other PRs! Feels like really good progress!!

[alanocallaghan]

cars is in episode 2, in which we've already created a linear model object, so should just call model.matrix on that instead of making a new one

[ailithewing]

These all look great. I think it's worth keeping the small number of datasets we have rather than using methylation for all. It illustrates the generalisability in approach and also gives biologists (our target audience) more applications to relate to without being unmanageable.