= ChimeraX Developer Meeting =

Attendance: Zach, Tom, Eric

February 5, 2024

= Agenda =

* Fix Chimera Windows build broken by updating cygwin.
* Update Cython and other prereqs.
* What ChimeraX 1.8 work/tickets should we prioritize for January?
* Worm ribbon implementation, ticket #9942
* Log slow to update when it gets long, Tristan ticket #9909
* Create Grafana dashboards to monitor web services.
* Make Chimera end-of-life bug report and registration changes for 1.18 release in Spring 2024.
* Back up the Chimera build machine to a solid state drive, Windows 7.
* ChimeraX Ideas milestone clean-up.

--------------------------
= Discussion =

* Windows Chimera build charge calculation
  - Eric fixed copying of fortran library to distribution.
  - Charge calculation on 2gbp worked.

* Chimera on Nvidia 4070
  - On vive.cgl.ucsf.edu Windows 10 showing only black graphics pane.
  - Replaced Quadro P6000 with Nvidia 4070 a couple weeks ago (for VR NIAID testing).
  - Chimera problem was it was using stereo mode.
  - Fixed by switching to mono camera.

* Speech to text for ChimeraX control
  - Scooter asked Zach if we can demo speech to ChimeraX commands during NIAID visit.
  - Zach trying OpenAI Whisper and a Google speech to text service.
  - Zach says you have to yell to get reliable speech to text.

* AlphaFold group meeting for John Gross lab
  - Tom giving presentation on running alphafold to find protein-protein interactions tomorrow.
  - [https://www.rbvi.ucsf.edu/chimerax/data/afbatch-jan2024/rim_dimers.html Slides]
  
* SPICE idatm typing
  - Eric talked to Peter Eastman about extremely unlikely protonation states in SPICE database.
  - Peter says as long as they are physically possible they are not considered a problem.
  - Eric will simply drop SPICE entries that cannot be idatm typed.
  - Will keep ones with exotic protonation.
  - Goal is to provide these idatm types to train a ML network to predict charges.

* Grafana
  - Zach says Scooter suggests running it as nightly cron job.
  - Last try to setup realtime monitoring generated lots of processes that Scooter had to kill, and Zach says all data files were erased.

* mmseqs2 sequence alignment on plato
  - Tom will try installing mmseqs2 sequence alignment for ColabFold on a plato node.
  - May use new SSD drive for databases.
  - Testing to see if ColabFold server will work with Rocky 8.
  - If it works we may get 1 Tb memory to run a ColabFold sequence server.


--------------------------
= Action Items =