2025.01 Release Notes
ASKCOS Development
Software package for the prediction of feasible synthetic routes towards a desired compound and associated tasks related to synthesis planning. Originally developed under the DARPA Make-It program and now being developed under the MLPDS Consortium.
askcos2_core
- Logging error resulting in frontend authentication failure on EKS instance (Issue askcos2_core#150)
- Template reactions are not found using api/reactions/search-reaction-id (Issue askcos2_core#149)
- RXNMapper dying (Issue askcos2_core#146)
- Drawing Bug for Prozac (Issue askcos2_core#144)
- Method Training_Set "USPTO_FULL" not working in the retrosynthesis prediction module! (Issue askcos2_core#140)
- added wrapper for fastsolv with config update (MR askcos2_core!115)
- add exception handling if logs_collection is uninitialized (MR askcos2_core!114)
- add data, scripts and README for FDA case study (MR askcos2_core!113)
- add sample script for async mcts (MR askcos2_core!112)
- propergate termination criteria properly; build_tree_options now getting... (MR askcos2_core!111)
- Rdkit draw (MR askcos2_core!110)
- Refactor/rework backup restore to use mongo cli (MR askcos2_core!109)
askcos-vue-nginx
- Safari Invalid Date Bug (Issue askcos-vue-nginx#359)
- Completed multiple fixes (MR askcos-vue-nginx!229)
- Changes for recent history on frontpage (MR askcos-vue-nginx!228)
tree_search/expand_one
- scscore ranking in ascending order (MR tree_search/expand_one!5)
- Sorting by SCScore should be in ascending order (Issue tree_search/expand_one#1)
retro_star
- turn off path shuffling (MR tree_search/retro_star!5)
tree_search/mcts
- turn off path shuffling (MR tree_search/mcts!5)
retro/template_relevance
- Move intra_only and dimer_only assignments inside the try/except block (MR retro/template_relevance!90
- GPU docker image is not updated (Issue retro/template_relevance#10)
rxnmapper
- add minibatching and hardcode batch size to 32 for long input(s) (MR atom_map/rxnmapper!3)
Deployment
We currently support two methods for deploying ASKCOS: Docker Compose (recommended) and Kubernetes (limited support). Docker Compose is a simpler method for deploying on a single workstation, while Kubernetes is more complex but is suitable for scaling across multiple nodes.
Hardware and Software Requirements
Please review the hardware and software requirements before proceeding to install ASKCOSv2.
Deployment options
ASKCOS is highly modularized and can be installed locally in two ways (click the links for installation instructions):
Upgrade Information
ASKCOSv2 can be easily upgraded by following the instructions here.
Member companies can find more deployment options on the members site