UNC at Chapel Hill – Machine Learning Via Web Application Bristol-Myers Squibb – Combinatorial Chemical Library Design / HTS Data Analysis & Visualization Amgen – Discovery & Preclinical Data Analysis & Visualization CHDI – Medicinal Chemistry CRO Management
NOW
CIMPLRX is discovering and developing new drug candidates using its own Explainable Artificial Intelligence (XAI) platform, CEEK-CURE.
CIMPLRX CAN PROVIDE INSIGHTS INTO DRUG DISCOVERY
Concept of XAI
Hover your mouse pointer over the image to observe the difference.
Platform
Target Based Hit Identification
A Uniprot ID is used to collect known compounds associated with the target and perform supervised and unsupervised learning. Generated models are used to design new compounds and screen commercial databases automatically.
Input: UniProt ID
Design new compounds and screen commercial libraries
Screen compounds in public databases
Generate directional models
Design new compounds and screen commercial libraries
Collect known relationships and build machine learning models
Compound based Hit Identification
Seed compounds are used to define the biological spaces and collect known relationships between compounds and targets. Generated models are used to design new compounds and screen commercial databases automatically.
Input: Seed Compounds
Collect data using 6 different chemical spaces
Build machine learning models, make predictions, and design new compounds automatically
Build directional models, make predictions, and design new compounds automatically
Lead Optimization
Prebuilt models are available to make predictions, and the results can be used to filter compounds. The color and size of circles shown on each structure are used to explain relative importance. Green and red colors indicate positive and negative contributions, respectively. The size of circles represents the strength of each contribution.
Color contributions
Filter results
moa identification
Active compounds are used to collect known compound-target relationships which are then turned into multiple predictive models. Top 3 targets per each compound and the occurrence of each target are shown.
Input: active compounds
Model generation and make predictions
Identify top 3 targets per each input compound and calculate their frequencies
Make comparisons between input compounds and compounds found in the training set
Explainable AI (XAI) Platform, CEEK-CURE
Explain Models
CEEK-CURE allows users to choose models and make predictions on a dataset of drugs. Results are shown with colored structures, prediction values, and standard deviations. Users can also combine different models for easy analysis.
Explain Results
CEEK-CURE provides various tools to explain results, including parallel coordinates, chord diagrams, dendrograms, heatmaps, interactive scatter plots, and clustering and diversity tools for structure-activity relationship analysis.
Explain Processes
CEEK-CURE displays the drug development process as a graph, allowing users to find related information through protein nodes or keyword searches. Active sites, compounds, and interaction information can be accessed from the Uniprot website. PyMOL can be used to view the protein’s active site, and protein structures, compounds, and data are automatically linked.
And Share
History
– 2 Government R&D Grants (Government Ministry of Health and Welfare & Ministry of Science and ICT / Ministry of Food and Drug Safety)
– Series B Funding (KRW 5.5BN (≈USD 5M))
– 1 Government R&D Grant (Korea Drug Development Fund, KDDF)
– Government R&D Grant (AI-Based Innovative New Drug Discovery)
– 2 Projects with Pharma (Dong-A ST / SK Chemicals)
– 1 Project with Biotech (Curex)
– Series A Funding (KRW 5BN (≈USD 4.4M))
– 2 Government R&D Grants (AI Drug Discovery Platform)
– Ulsan Univ. Project
– Moved into AI YANGJAE HUB, SEOUL
– CIMPLRX CO., LTD. Established
Our Partners
University
Biotech & Pharmaceutical Companies
Investment Institutions
Business Model
contact
02-6427-3100 / info@cimplrx.com 8F, AJ Building, 9, Jeongui-ro 8-gil, Songpa-gu, Seoul, Korea
about us
CIMPLRX is discovering and developing new drug candidates using its own Explainable Artificial Intelligence (XAI) platform, CEEK-CURE. Create ExplainablE Knowledge (CEEK) is the user interface portion of the platform designed to analyze and visualize results and capture the discovery process. Collect and Uncover RElationships (CURE) is our XAI engine which resides on the server side and handles various requests made from CEEK.
