Recent publications impressively demonstrate the benefits of our solution.
Challenge with false hits​
„… 92,6 % of drug targets identified in preclinical studies are false positives, because the causal connections between genes or proteins and a disease (MoA) were not correctly identified.“
Distinction between PDD and TBDD

„…Approaches to drug discovery can be broadly divided into two categories: PDD (phenotypic drug discovery) and TBDD (target-based drug discovery).“
„…PDD approaches are more successful in discovering new first-in-class drugs than TDD.“

Higher classification accuracy through PDD with Convolutional Neural Networks (CNN)

„…Compared to ML based on manually crafted features, CNNs improved the classification accuracy by approximately 10% in PDD (cell-based phenotypic drug discovery) The CNN was employed for drug screening and (…) it was demonstrated how CNNs can be used for robust hit identification in HCS without fluorescent labeling.“

PDD and the mode of action (MoA)

„…An important goal in drug discovery is determing the MoA of drug candidates. An approach to achieve this imaging and ML (PDD (cell-based phenotypic drug discovery) has been successfully demonstrated for antibiotics. Their approach was used to predict the previously unknown MoA of a new drug, and the predictionn was experimentally calidated very well. “

Conclusion from the “Phenotypic Drug Discovery” study

„… (PDD) will undoubtedly accelerate progress in drug discovery, significantly speeding up the identification of novel MoAs and novel drugs, reducing R&D costs, and giving PDD new momentum.“
You can find the full publication here


Our PDD module (cell-based phenotypic drug discovery) is currently being applied in three core applications.

Classic - We take over the cell screening for your active ingredients. Fast, efficient, and highly automated

High-Content-Screening (HCS) is a microscopy-based high-throughput screening (HTS) using an advanced Cell Painting PDD approach: six dyes to stain eight cellular components and organelles (nucleus, endoplasmic reticulum, nucleoli and RNA, F-actin cytoskeleton, Golgi, plasma membrane, and mitochondria), enabling a more detailed characterization of cellular morphology
Example application: Immunology

Cells: Macrophagen (M0, M1, M2, Dead Cells)
References: Leading German pharamaceutical company, Pasteur Institute Paris.

Cells: Macrophagen (M0, M1, M2, Dead Cells) References: Leading German pharamaceutical company, Pasteur Institute Paris.

Combining microscopy-based high-throughput screening (HTS) with transcriptome data through multimodal clustering (integration of transcriptome and microscopy data). This enables the identification of drug hits without the need for genetic manipulation of cells while also gaining transcriptome information without generating these data.
Example application: Oncology

Cells: Cancer cells, Osteosarcoma cells
Reference: aimed analytics, ZIM funding program


Hybrid methodology for direct processing of 3D data such as tissue sections or similar. Combination of 3D CNNs with other techniques such as Graph Neural Networks (GNNs)
Example application: Dermatology, skin aging, skin diseases

Cells: Skin cells + cannabinoids
Reference: Fraunhofer IGB, 420 Pharma, ZIM/Eureka

Our PDD module (cell-based phenotypic drug discovery)

This model is currently under development. AI-based solution for the manufacturing process of biopharmaceuticals, which enables high-producer single cells to be identified and (semi-) quantitative statements to be made about the production capacity of the cells. The result: a significant reduction in the production time of biopharmaceuticals and a considerable reduction in material requirements
Example application: Biopharmaceuticals

Cells: High producer cells, production cell lines
Reference: Fraunhofer IGB


Are you interested in our solution or in collaborating? Contact us!


Discover the Future

Book a demo now to see how our platform accelerates research, reduces costs, and improves success rates through automated cell image analysis.

Screening Hub is a trademark of ValueData GmbH
Wankelstraße 14
70563 Stuttgart
+49 711 674 287 11

auf deutsche Version wechseln.