DC3

AN INNOVATIVE DATA-DRIVEN DRUG DISCOVERY PLATFORM FOR ALZHEIMER’S DISEASE

 

Home institution: Nostrum Biodiscovery – Spain (24 months)

Host institution: Alma Mater Studiorum University of Bologna - Italy (10 months)

PhD Enrolment #1: Universitat de Barcelona – Spain

PhD Enrolment #2: Alma Mater Studiorum University of Bologna - Italy

Supervisor #1: Lucía Díaz

Supervisor #2: Barbara Monti

Intersectoral Secondment: not needed

Objectives: NBD is a biotech company working together with several major pharmaceutical companies in Europe and the US and has a wide knowledge of the requirements of pharmaceutical companies working in rational drug design. NBD advantages are its technologies, allowing the saving of costs and the acceleration of the whole process. Secondly, the level of calculation and prediction capacity by computation, which directly impacts research reliability. Lastly, the team experience and the research excellence which has been strongly demonstrated by the internationally recognized institutions that support this project for years now.

The doctoral candidate will work for two years in Spain (Nostrum Biodiscovery) and for 1 year in Italy (University of Bologna).

The ADDP platform will be developed by using as a starting point the tools existing at NBD for compound database profiling and de novo compound generation. The work will be developed recursively in two parallel lines of work that will serve as ground-breaking use of AI methodologies and as a baseline to compare and improve the final outcome. The baseline will be executed as follows: (1) data assembly and curation; (2) 2D featurization of billion compound libraries and known inhibitors for targets of interest; (3) feature matching; (4) in silico validation using PELE Platform/PELE-AI. In parallel, development and appliance of generative model architectures such as VAEs and GANs will be integrated in the following pipeline: (1) data assembly and curation; (2) de novo compound generations in a low data regime; (3) 2D featurization; (4) feature matching to compound libraries; (5) in silico validation using PELE Platform/PeleAI of repurposed and generated compounds. This workflow will allow to expand the known chemical space for finding and repurpose drugs targeting molecules involved with AD. The repurposed molecules will be purchased by available vendors and preliminary screened for their potential toxicity in CNS cell models. Then non-toxic molecules will be progressed for proof-of-concept biological studies, including assessing drug synergy in combination studies. In addition, an initial PK characterization will be performed experimentally in collaboration with DC5.

Expected outcomes: The fellow is expected to generate breakthrough ideas in the assigned area of research, as well as to carry out research in line with the project plan. The fellow is also expected to contribute to dissemination and communication activities, besides actively collaborating within the TClock4AD network. This doctoral candidate will work closely with consortium partners at University of Bologna.