Our approach
Leaning on the power of AI, we shift the drug discovery process into patient-based cellular models.
Our versatile deeptech drug discovery platform exploits outstanding expertise in human cellular disease modeling, assay automation and validation, together with key assets in AI-powered phenotypic profiling and data integration to reveal novel therapeutic opportunities.
We functionalize patient-based disease models for Target and Drug Discovery
AI-assisted high-content analysis renders complex disease models applicable for large-scale phenotypic target and drug discovery.
PREDICTIVE.
Complex patient-based models. AI allows us to tackle intrinsic patient cells batch-to batch variability and to establish authentic and predictive disease models.
VALIDATED.
Validated Assays. Proof-of-concept achieved as clinically effective drugs rescue Ksilink AI-driven morphometric readout of patient-based models in assay format.
SCALABLE.
“Scalability thinking” from the start. Our Image & Data mining team develops tailored algorithms to deal with large amounts of biological data generated in house from automated assays.
INTERPRETABLE.
Tailored algorithms. Our team develops tailored algorithms to extract information from large datasets generated in house and enrich / corroborate with external data.
Driven by AI
The power of AI is key for all our activities. We use Profiling and Deep Learning methodologies to analyse thousands of subtle morphometric read-outs, allowing for the compensation of inherent variability of patient-based disease models. AI also enables fast interpretation of our generated imaging datasets, deciphering human disease biology.
Our focus on phenotypic profiling of complex patient-based assays sets us apart in the field.
Assay Development & Validation
We create patient-based cellular assays with a focus on robustness and high-throughput scalability to conduct large scale phenotypic screening, using genetic modifiers like siRNA, CRISPR, or small molecular compound libraries. From the start we focus on phenotypic readouts using microscopic imaging. To ensure the relevance of our findings, we customize combinations of biochemical tests, functional readouts and disease specific tool compounds for a final pharmacological validation.
Our focus on scalability and validation sets us apart in the field.
Image & Data Mining
Data is at the heart of what we do at Ksilink. Critically, dedicated AI algorithms (see movie on the right) allows the fast and precise classification of compounds for their ability to shift diseased-like phenotypes (DCM / WID-B ) towards the healthy control (WT). CRISPR screens even allow for a simultaneous identification of relevant compounds and their corresponding targets. Knowledge graphs are used for dataset integration and decision finding to select most relevant targets and candidates.
High-Throughput Phenotypic Screening
Our sterile robotic screening facility allows high-throughput, image-based screening of patient-based assays over extended periods. In parallel, Ksilink has secured access to large scale, quality controlled and standardized production of differentiated, patient iPSC-based material. Together, it allows for high-throughput screening and AI boosted analyses directly in the relevant disease model, performing the discovery process in a “human proof-of-concept” setting.
Our unique capacity ensures that our findings are both comprehensive and disease relevant.