-
Notifications
You must be signed in to change notification settings - Fork 0
/
meeting_saved_chat_July24.txt
100 lines (51 loc) · 4.78 KB
/
meeting_saved_chat_July24.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
12:01:09 From Harvey Lieberman to Everyone:
R/Pharma etiquette. Please be polite to all participants. Mute microphone unless speaking.
R/Pharma hangouts are not recorded so please speak freely. We understand that thoughts are those of individuals and not of the institutions that they represent.
Any issues please ping one of the co-hosts (Phil, Jared or Harvey).
gtsummary https://phuse.s3.eu-central-1.amazonaws.com/Archive/2024/Connect/US/Bethesda/POS_PP20.pdf
webr pilot submission https://rconsortium.github.io/submissions-wg/pilot4.html
https://quarto.org/docs/dashboards/
https://books.ropensci.org/targets/
duckdb + wasm sounds 🔥
dataset.json
https://www.cdisc.org/dataset-json
https://advance.phuse.global/display/WEL/Dataset-JSON+as+Alternative+Transport+Format+for+Regulatory+Submissions
R/Pharma dataset-JSON (Jesse Anderson, FDA) Dataset-JSON as Alternative Transport Format for Regulatory Submissions | R/Pharma (rinpharma.com)
For fun historical context, they started talking about this in 2017 which is probably why not parquet https://wiki.cdisc.org/download/attachments/124204045/Transport%20for%20the%20Next%20Generation%20Version%201.0.pdf?api=v2&modificationDate=1628184569292&version=1
Lots of good duckdb, parquet info here https://posit-conf-2023.github.io/arrow/
The 2024 workshop will have even more on those topics...once the authors finish writing it 😅
Xpt is also a binary
it would be great to hear more about everyone's set ups around reproducibility aspect of generating outputs in R (R packages, or transitions from rmarkdown to quarto etc., maintaining Shiny apps) not only on a submission level but also general daily work
Containers. It's the only way to be sure...
+1 to containers plus podman over docker because of FDA preference
podman yes
https://podman.io/
https://www.appsilon.com/post/docker-vs-podman-vs-singularity#key-differences
Singularity on HPC
Also completely BANNING desktop R... ;-)
https://rconsortium.github.io/submissions-wg/pilot4.html podman etc
Buf if you have a container - what information do you need to retain to reproduce the container? The whole container or just the Dockerfile or singularity recipe ? How tight do you need to be to get away with just the Dockerfile/singularity recipe
https://github.com/sol-eng/singularity-rstudio
I wrote some about these topics here https://www.lexjansen.com/phuse-us/2024/os/PAP_OS10.pdf
down in part 6. Images/HPC/Docker
https://wlandau.github.io/crew.aws.batch/
https://books.ropensci.org/targets/cloud-storage.html
https://www.nextflow.io/
submissions need to be open in general. software and data as much as possible
https://insightsengineering.github.io/nest/
For me, “pre-competitive space” is often the verbiage that does the trick with executives
novo https://www.youtube.com/watch?v=t33dS17QHuA&ab_channel=PositPBC Novo spoke about that here too:
https://phuse.s3.eu-central-1.amazonaws.com/Archive/2024/Connect/US/Bethesda/PRE_HoW01.pdf
Just as a Public Service Announcement: I am working through an R Consortium Working Group to take the "Master the Tidyverse" / "Remaster the tidyverse" course material and recast it using pharma data examples. Anyone interesting in collaborating, please drop me an email at [email protected] or follow the repo here: RConsortium/remaster-the-tidyverse-wg: Converting "Master the tidyverse" training material to use pharma-centric data. (github.com)
that sounds interesting!!
Would love to know how people deal with communicating package versioning in submissions - I've seen some reports of pushback on using things like renv whereas other folks have had positive responses to it
Speaking of Rshiny, is it possible to use open-source r package or R Shiny package for one's own service? For example, {teal} provides a really good template and can we manipulate those modules and create a Shiny apps based on that for other service?
GSK spoke about this some here https://www.lexjansen.com/phuse-us/2024/os/PAP_OS16.pdf
For me I am asked to train other pharma companies on my open source packages. This is often where I get push pack
would be awesome to get some of it here too https://pharmaverse.github.io/examples/
Feel free to ping Shannon Pileggi on the conversion process as I know she had to do the keynote -> quarto jump for the WTF workshop materials from Jenny
I'm actually hoping that we can use similar example datasets across pharmaverse / admiral and Master the Tidyverse so that there's a bit of continuity. So come and learn the tidyverse way, then let's show you the pharmaverse way...
https://rinpharma.com/contact/ [email protected]
Thanks - this was excellent, really enjoyed participating, thanks folks for organising :)