Minutes 2021-12-03

Author

Joseph Rickert

Published

December 3, 2021

  • “Michael Blanks - Beigene”
  • “Gregory Chem”
  • “Bryant Chen - FDA”
  • “Hye Soo Cho - FDA”
  • “Heidi Curinckz”
  • “Robert Devine -Johnsom & Johnson”
  • “Bob Engle - Biogen”
  • “Miram Fossati - Merck”
  • “Steven Haesendonckx”
  • “Ryan Johnson - RStudio”
  • “Ning Leng - Genentech”
  • “Tadeusz Lewandowski - Roche”
  • “Eli Miller - Atorus”
  • “Emily Nguyen - FDA”
  • “Joseph Rickert - R Consortium”
  • “Frederic Saad”
  • “Paul Schuette - FDA”
  • “Jonathan Tisack - Beigene”
  • “Bryan Tegomoh”
  • “Hao Wang - FDA”
  • “Heng Wang”
  • “Nan Xiao - Merck”
  • “Hong Yan - Regeneron”
  • “Renping Zhang - FDA”
  • “Jizu Shi”

The meeting was recorded and the video is available with passcode: m4Jfez!E

JBR opened the meeting and remarked that the announcement of the recent successful completion of the Pilot 1 submission generated quite a bit of activity on social media, and that he has received several requests from people who would like to join the working group, or at least attend some of the meetings. JBR stated that it is great benefit to the R Community and to the R Consortium that all interested folks can join working groups. There is no requirement that working members be employed by companies who are members of the R Consortium. He noted that, so far, having large numbers of people attend the monthly meeting of the submission working group meeting has not been a problem.

JBR also noted that small teams of WG members have been meeting separately from the general meeting to accomplish the technical work and asked that all of these please make an effort to take minutes and upload them to the appropriate repository. These meetings can be very brief limited to who attended and some bullet points of what was accomplished.

JBR then ask NL to give a brief overview of WG activity. NL stated that in addition to the Pilot 1 team, there are two groups working on projects.

  1. The Pilot 2 submission that is developing a shiny app based on the analyses developed in the Pilot 1 submission. This app will then be made into an R package, wrapped up with the analysis and data used in Pilot 1 and then submitted through the FDA gateway. The team working on this meets every two weeks and has already developed a preliminary version of the shiny app. They will ask the extended working group to review the draft shiny app. if all goes well they think they should be able to make the Pilot 2 submission in two to three months.

  2. Another subgroup has just been formed to make a Pilot 1 submission the PMDA, the Japanese health authority. This will essentially be the Pilot 1 submission to the FDA with some translate materials. SH and HW are are members of this team. TL asked to be included.

At 16:16 minutes into the video TL demonstrated the preliminary of the Pilot 2 Shiny App. He described how filtering and subsetting works.

 Opening page of preliminary Pilot 2 Shiny App

JBR suggested that eventually the shiny app should be deployed to an R Consortium property.

The group they began a discussion about how the Shiny App would fit into a typical FDA workflow. PS that his group has not yet dealt with any Shiny apps and they are only in the preliminary stages of developing their own apps. He also remarked that there would still be a significant amount of preliminary work to set up for an analysis.

PS said that an issue with Shiny will be reproducibility. He remarked that he is following the work of Joe Cheng and Carson Sievert on the Shiny Meta project which will generate code that captures the Shiny app settings to allow users to independently rerun an analysis.

PS remarked that the ability to do subgroup analysis would be a value add to the reviewers. The Shiny app would be an auxiliary tool. The main task would still be to do an independent analysis of the models and show that the models are appropriately developed and executed.

RJ then asked PS whether the FDA has the infrastructure to receive a Shiny app and deploy it throughout the enterprise. PS replied that do not yet have the capabilities for enterprise deployment, but that two different possible solutions are being discussed: one of these is OCS the Office of Computational Science.

Right now, the FDA is running things locally. The best that the FDA can do is to install a shiny app and submission package on a number of workstations which would provide remote access to FDA analysts. The workstations would not be networked. At the present time, the FDA does not permit analysts to use servers.

JBR asked PS: “Given the FDA environment as it is now, not as we all would like it to be, is there anything the we can do do facilitate your workflows?”

PS replied that the FDA is in the process of updating its workflows including the use of GitLab, but that he does not know how they could make things easier without achieving the enterprise environment being discussed. JBR remarked that if any of the FDA

JBR if anyone in the meeting was working on the containers project. EM stated that he would like to be involved. NL noted that Eric Nantz has already begun collecting documentation. JBR pointed out that researching the use of containers for R submissions would be a project that newcomers to the Submissions WG could help with.

In summary, the WG has two active project and two that are just beginning: 1. The Shiny app for the Pilot 2 submission 2. Research on containers for Pilot 3 3. A Pilot 1 submission for Japan 4. A pilot 1 submission for China

The next meeting of the WG will be on Friday May 6, 2022 at 9AM Pacific Time.