Project (Kerstin Lenhof)

SenDoCo: Dose-specific Drug Combination Sensitivity Prediction

For a variety of complex diseases, the administration of drugs or drug combinations is a major treatment modality, e.g., cancer [1], infectious diseases [2], depression [3], or cardiovascular diseases [4]. However, developing machine learning (ML) models that match individual patients to the most effective drug treatment while also minimizing unwanted side effects has proven to be extremely difficult: compared to the complexity of (human) biology, data are scarce, the acquisition of pharmacogenomic data from patients is still intractable on a large scale, and the exact mode of action of many drugs remains to be elucidated.
Within this PhD project, we aim to combine a holistic systems-biology perspective with cutting-edge ML methods to address this task in anti-cancer drug responses. Briefly, we take a multi-omics multi-disease perspective to inform biologically inspired neural networks (BINNs). The multi-disease perspective should help alleviate data scarcity issues while also promoting generalization across tasks, possibly allowing for the application of the developed method in other domains. The integration of a priori knowledge within BINNs should ensure that current biology is adequately represented, which should not only help overcome data scarcity but also simultaneously increase the interpretability, which is thought to be a major supporting factor for achieving human oversight
for ML in the medical domain [5].
The project offers a unique opportunity for a PhD student to work at the intersection of computational biology, ML, and medicine, contributing to both methodological development and translational efforts within the clinic. The PhD student will also be embedded in a vibrant research environment at the University Medical Center G¨ottingen (Medical Bioinformatics and Oncology department) and benefit from close ties to the Lower Saxony Center for AI and Causal Methods in Medicine (CAIMed).

Tasks
• advancing of dose-specific drug combination sensitivity prediction
• developing multi-omics multi-disease biologically inspired neural network (BINNs)
• identifying novel biomarkers of drug response that should be experimentally validated within the University Medical Center G¨ottingen
• advancing transfer learning strategies to enable deployment in the real-world

Your interests
• you like programming
• you like organizing and structuring knowledge
• you want to use (computer) science to foster societal well-being

Your skills
• solid programming skills in python
• knowledge of the foundations of machine learning, particularly deep learning
• solid knowledge of biology (knowledge of the biology of cancer would be advantageous)

Useful resources to get started
To obtain an initial impression of the work in our group related to this proposal, consider reading [6, 7, 8, 9, 10]. Please visit our group website (https://bioinformatics.umg.eu/research/workgroups/ki-and-multi-omics-data/) to get more information about members and ongoing projects of the group!






Homepage Research Group
https://bioinformatics.umg.eu/research/workgroups/ki-and-multi-omics-data/



Publications:

[1] Vasan, N., Baselga, J., & Hyman, D. M. (2019). A view on drug resistance in cancer.
Nature, 575(7782), 299-309.

[2] Oselusi, S. O., Dube, P., Odugbemi, A. I., Akinyede, K. A., Ilori, T. L., Egieyeh, E., ... & Egieyeh, S. A. (2024). The role and potential of computer-aided drug discovery strategies in the discovery of novel antimicrobials. Computers in biology and medicine,
169, 107927.

[3] Lee, S., Mun, S., Lee, J., & Kang, H. G. (2024). Discovery and validation of protein biomarkers for monitoring the effectiveness of drug treatment for major depressive disorder. Journal of Psychiatric Research, 169, 7-13.

[4] Ingelman-Sundberg, M., & Pirmohamed, M. (2024). Precision medicine in cardiovascular therapeutics: Evaluating the role of pharmacogenetic analysis prior to drug treatment. Journal of internal medicine, 295(5), 583-598.

[5] Sterz, S., Baum, K., Biewer, S., Hermanns, H., Lauber-R¨onsberg, A., Meinel, P., & Langer, M. (2024). On the Quest for Effectiveness in Human Oversight: Interdisciplinary. Perspectives. The 2024 ACM Conference on Fairness, Accountability, and Transparency.

[6] Lenhof, K., Eckhart, L., Rolli, L.M., Volkamer, A., & Lenhof, H. P. (2024). Reliable anti-cancer drug sensitivity prediction and prioritization. . Scientific Reports, 14, 12303. https://doi.org/10.1038/s41598-024-62956-6

[7] Eckhart, L., Lenhof, K., Rolli, L.M., & Lenhof, H.P. (2024). A comprehensive benchmarking of machine learning algorithms and dimensionality reduction methods for drug sensitivity prediction. Briefings in Bioinformatics, Volume 25, Issue 4, bbae242, https://doi.org/10.1093/bib/bbae242

[8] Lenhof, K., Eckhart, L., Rolli, L. M., & Lenhof, H. P. (2024). Trust me if you can: a survey on reliability and interpretability of machine learning approaches for drug sensitivity prediction in cancer. Briefings in Bioinformatics, 25(5), bbae379. https: //doi.org/10.1093/bib/bbae379

[9] Eckhart, L., Lenhof, K., Herrmann, L., Rolli, L. M., & Lenhof, H. P. (2025). How to predict effective drug combinations–moving beyond synergy scores. iScience, 28(6). https://doi.org/10.1016/j.isci.2025.112622

[10] Rolli, L. M., Eckhart, L., Volkamer, A., Lenhof, H. P., & Lenhof, K. (2025). Increasing trustworthiness of machine learning-based drug sensitivity prediction with a multivariate random forest approach. ChemRxiv preprint. https://doi.org/10.26434/chemrxiv-2025-ml78s