softloud/spunk

Nutritional interventions for male infertility: a systematic review and meta-analysis

spunk

spunk's Issues

update ms

Ranking interpretations:

• count
• volume
• motility
• morphology

next

• comments email

also

• pregnancy
• link to whatsit
• clear emails
• include discussion of sucra
• include links to matrices

reproducible contributors

Also I haven't had time to properly read the literature on what label is meant to be what. I just stick things in until it doesn't complain.

wrap facet text and rotate sensitivity

impromptu eshackathon (BMJ submission) or whenever (living analysis)

heya evisynth peeps, @DrMattG, @nealhaddaway, @mjwestgate, @befriendabacterium, @dmphillippo and opensci peeps, @yochannah, @ljcolling, @debruine, @steele (Lisa and James have rockstar handles) thought you might be interested in trying out living analysis collaboration workflows with me. I can email you a link to the manuscript whenever you decide to get involved.

I've put together first thoughts on workflow here.

Whenever suits.

arse backwards

Things are moving fast with data scaling up. Traditional diagnostic methods for statistics are less useful at combinatorially-many comparisons. And reproducibility is widely viewed as a end of production component. I was given free rein on this project, and because of that, I think I did better science than if I had attempted to implement fancy models. I followed the reasoning and guidance I'd been given before, but this was a road to frustration. Decided to go in a different direction, seeing as PI is letting me take the lead on this. Because there is so little time to get an analysis done in evidence synthesis. This seems more useful for the time allocated, if it makes it easy for others to also volunteer their time.

I am sorry to not give you more time to be part of this Cochrane review. This is why I am going to (wait, no, I wrote that bit already, almost done with draft) write in the methods section about why I think making an analysis patchable and provide shortcuts for common analyses in the discipline, that this should be done before implementing more advanced models. This is a phase one of an analysis, setting it up so that it's really easy to get help from the community, because science is better when collaborative.

An Australian might say implementing reproducibility and living analysis after the advanced models is done is arse backwards. Collaborative data science brings it's own challenges, but will absolutely help with the advanced algorithms.

Cochrane submission hackathon

I am pushing hard to wrap my component of the writing on the manuscript right now, I'm very behind because anxiety problems on top of the methodological concerns.

But I imagine some work will be done by the postdoc leading the paper, and the PI. So, if you see an issue you can do easily right now, your name will go into the Cochrane submission. And hey, there'll be revisions, too, most likely. So there'll be some weeks before final submission, surely. I'll update when I know more.

But also, whenever

It is a living analysis, however, and I don't see why major extensions aren't publishable. Makes it more useful, the peeps would want to know. I'm in this project to tinker with minimal workflows for this kind of thing, so publishing is a secondary concern for me. Happy to work on big and small aspects of this.

where has Charles been?

Long time mostly I've been quiet on the research front for a while. Feels like years since I've been able to think clearly. This happened in my twenties, but not for such a prolonged time since. I stopped speaking to everyone for about six months, trauma-induced anxiety at such a deafening roar. Suffice to say Mr Robot is basically porn for me at the moment.

I moved to Copenhagen for a job at a start up a few weeks ago, but they are okay with me investing a little time in this project. Life is better than ever here. I finally was in a good enough place to get some science done on spunk::, a side project I've had going (anxiety-induced procrastination) for about a year. PI has been so amazing and understanding every step of the way. I couldn't let a grrl down, or her postdoc.

wider figures visnetwork and sensitivity

transitivity

• investigate what else means a transitivity violation, in addition to too long a chain
• a place to start for explanations

refactor levels in sensitivity analysis so NMA is lowest

interventions in comparison matrices

league tables for the swame interventions as we have in network plots

fix site

• build
• use png

round rank means to integers

remove personal gh pkgs

• remove dontpanic
• remove pandan

task 1 finalise network diagrams and tables with text formatting

• Apply str_to_sentence to interventions
• Vitamin names as capitals; words with 1 word letter and up to two numbers capitals
• One size smaller
• the font is not cut off on the right hand side.
• If possible, make it one size smaller so it's not so crowded.
• zinc/zinc + folic acid should be Zinc/Zinc + folic acid
• vitamin/mineral combos should be Vitamin/mineral comb.
• vitamin d3 should be Vitamin D3vitamin c/e should be Vitamin C/Espirulina should be Spirulina. You get the picture. The first letter of each group should be capitalised, and the vitamin names (D, E, Q10, C etc) should be capitals too.
• Make tables with same capitalisation

explanation of MA, MA-1, and NMA in caption of sensitivity plot

port comparison tables to function

• link to img on gh in ms
• remove from site for now

Would be good to have helper function to get smaller subset of comparison tables. Comparison tables are currently in targets. *

So, would need, for starters.

• append class_1 and class_2 labels
• port gt code to function
• replace target code with function
• put subset comparison tables at top of comp.Rmd with code

sucra

• tried current plot
• see if rank raw dat is already extracted
• produce posterior rank probabilities
• new ggplot, perhaps use class
• x-axis less ticks
• outcome title
• less facet columns, 3 or 4