Within the AIDD consortium, there is an enormous need for computational approaches guiding the prioritization of compounds, experiments and research directions. As the lead group of the Christian Doppler Laboratory for Molecular Informatics in the Biosciences (CDLab-MIB), the Kirchmair team has a long-standing interest in developing and applying computational methods for predicting the biological activities, ADME and toxicological properties of small molecules, with focus on natural products and anti-infective agents (Chen 2020). The AIDD-team can build on ample structural information available on bromodomain and extra-terminal domain (BET) proteins, as well as information on lectins LecA and LecB, and the outer membrane transporters for siderophores (PfvA, FptA and CntO). Moreover, the team can draw from information on known fragment-sized and small-molecule modulators of the proteins of interest.

The Kirchmair group will integrate state-of-the-art computational approaches to develop an in silico platform tailored to the identification and optimization of fragment-sized and small-molecule modulators for the anti-infective targets of interest to the AIDD consortium. The virtual screening module will feature a structure-based module, where docking and pharmacophore models will be used to screen molecules against the accessible experimental structures and homology models derived and validated with state-of-the -art methods. The platform will also include a ligand-based module utilizing pharmacophore-based and shape-based screening approaches. The research focus of the PhD studies will be the investigation and development of methods for maximizing the applicability of machine learning approaches to molecular property prediction for natural products (most of the available data and, hence, available models are focused on synthetic compounds). Specifically, the PhD student will investigate different types of knowledge-sharing approaches in machine learning (as exemplified in recent work from the group, Wiercioch 2021), with the aim to boost predictions for the natural products chemical space, where experimental data are scarce. In terms of applications and case studies, particular interest will be on exploring bioactivity, ADME and toxicity predictors for natural products to guide the experimentalist groups of the consortium. In addition, the CDLab-MIB will provide their data warehousing infrastructure and digital tools to the AIDD consortium.