Informatics Matters Overview

Owned by Skyner, Rachael (DLSLtd,RAL,LSCI) (Unlicensed)
Last updated: Nov 15, 2023 by Max Winokan
2 min read
20-Jan202027-Jan03-Feb10-Feb17-Feb24-Feb02-Mar09-Mar16-Mar23-Mar30-Mar06-Apr13-Apr20-Apr27-Apr04-May11-May18-May25-May01-Jun08-Jun15-Jun22-Jun29-Jun06-Jul13-Jul20-Jul27-Jul03-Aug10-Aug17-Aug24-Aug31-Aug07-Sep14-Sep21-Sep28-Sep05-Oct12-Oct19-Oct26-Oct02-Nov09-Nov16-Nov23-Nov30-Nov07-Dec14-Dec21-DecMarker
Deployment
Workflows infrastructure
JackFS

Fragalysis stack to kubernetes

Fragalysis stack CI/CD

Fragment network to kubernetes

Fragment network CI/CD

Squonk to galaxy link

Squonk to fragalysis link

Scoring containerised

Finalise docking workflow

Scoring tool in galaxy

Workflow in galaxy

Actions:

Share time spreadsheets (TD)
Create github repo for fragalysis documentation (RS)
Send Rachael a list of all github repos (& context) (TD)

Major tasks:

  • Fragalysis stack → kubernetes prototype

    • Backend neo4j

    • Deploy JupyterHub (ORN version)

    • Switch to postgres

    • Fragalysis backend to Fragment Network by API

  • Fragment network → kubernetes DONE

  • Squonk → kubernetes DONE

  • Galaxy → kubernetes planning

  • Migrating Verne to K8s Planning

  • Fragalysis stack CI/CD PENDING

  • Fragment network CI/CD PENDING

  • Openshift K8s HOLD

Notes:

  • Early March for K8S

  • Don’t need CI/CD for demo on kubernetes – should we hold off for 2/3 months? (Travis vs. Jenkins)

  • Testing of documentation (Janssen)

  • Process for AWS exists, but openstack doesn’t (for DLS) – 1 week after start (as long as networking, accessibility etc. work)

  • Need to discuss Galaxy:

    • use helm charts instead of playbooks?

  • Fragalysis questions: Mechanisms for running the data loaders?

  • Janssen: Scott will have a better idea of wether it will be applicable outside of the Fragalysis scope/use (still investigate)

  • Pin down date for starting deployment

  • Long TC for deployment on AWS

  • M2M/Janssen can deploy - can run playbook, entering version that deploys stack to cluster

 

Actions:

Follow-up on technology stack questions with Janssen (TD)
Second technology stack meeting - arrange (RS) [week after next]
Send containers to Janssen so they can run through tests (TD) - keep sending
Link Janssen up with Björn for Galaxy deployment
Documentation Repo (for deployment) (RS)
Dates for STFC for deployment (openstack) (TD/AC)

Major tasks:

  • Squonk to galaxy link unknown

  • Fragalysis to squonk link unknown

Actions:

Review timeline/roadmap

Major tasks:

  1. Scoring method containerised DONE

  2. Finalise docking workflow in progress

  3. Insert Bayesian optimisation discovery

  4. Implement scoring into Galaxy tool unknown

  5. Docking+scoring workflow in galaxy unknown

  6. Improve container HOLD

Notes:

  • Currently putting full workflow together before it goes into squonk/galaxy

  • Orchestration between layers needs thought

  • Options should be configurable (e.g. number of poses to dock)

Actions:

Chat to Galaxy guys to include GPU node and configure to target it (TD)
Check current docking workflow (RS) – arrange f2f @SGC meeting for Friday afternoon (RS)
Scope out enumeration step – should we add as galaxy tool? (TD)
Scores and poses into SDF file as output of scoring step (TD)

Major tasks:

  1. Improve fragmentation algorithm in progress

  2. Define data-building workflow in progress

  3. Implement API HOLD

  4. Add compounds into network via. API IN progress

Notes:

  • Relational database - being used to index molecules (all datasets that are processed) – it’s where the molecules are stored before the graph network is created

  • Upper limit (guess) of adding molecules directly into network is ~10,000

 

 

Â