Machine learning software to design personalized neoantigen vaccines tailored to specific vaccine delivery systems

Información del proyecto

MEDIVAC

Identificador del acuerdo de subvención: 775611

DOI

10.3030/775611

Proyecto cerrado

Fecha de la firma de la CE 6 Mayo 2017

Fecha de inicio 1 Junio 2017

Fecha de finalización 31 Octubre 2017

Financiado con arreglo a

• SOCIETAL CHALLENGES - Health, demographic change and well-being

Coste total

€ 71 429,00

Aportación de la UE

€ 50 000,00

Coordinado por

NEC ONCOIMMUNITY AS
Norway

Periodic Reporting for period 1 - MEDIVAC (Machine learning software to design personalized neoantigen vaccines tailored to specific vaccine delivery systems)

Período documentado: 2017-06-01 hasta 2017-10-31

Resumen del contexto y de los objetivos generales del proyecto

Cancer remains a leading cause of disease and death worldwide. An estimated 14m new cancer cases were diagnosed in 2012 and the annual cost of treating cancer in the EU is estimated to be over €150b. Thus, there is a clear need for more effective methods of treatment to reduce this global burden. Immunotherapy, and particularly personalized cancer vaccines represent an exciting new weapon in the war against cancer. However, to develop personalized cancer vaccines, it is necessary to identify personalized tumor-specific targets known as neoantigens in a cost effective and clinically actionable timeframe.

OncoImmunity is currently developing a state-of-the-art software solution that addresses this problem– the Immune Profiler prediction engine (PE) which leverages advanced machine-learning algorithms and bioinformatics approaches to predict whether a potential neoantigen has the correct physiochemical features to be immunogenic. However, to fully enable the development of personalized vaccines this software need to be customized to specific vaccine delivery systems (VDS) as different vaccine platforms induce responses against different repertoires of neoantigen. The tailored software will be called MEDIVAC.

Trabajo realizado desde el comienzo del proyecto hasta el final del período abarcado por el informe y los principales resultados hasta la fecha

To develop MEDIVAC the core Immune Profiler machine learning (ML) framework needs to be adapted to enable the client to iteratively train the framework on their own clinically validated data to generate VDS specific predictions. In addition, the software must be able to optimize vaccine constructs to minimize off-target autoimmune responses. The work performed in this project assessed the feasibility of adapting the core Immune Profiler PE for this purpose. In addition, the project mapped out the pathway for combining these novel features in a validated and CE certified software package that will enable OncoImmunity to take a leading position in this rapidly growing market. The conclusion is that the development of the MEDIVAC software is feasible, and that the software can be developed, and CE marked as an in vitro medical device.

Avances que van más allá del estado de la técnica e impacto potencial esperado (incluida la repercusión socioeconómica y las implicaciones sociales más amplias del proyecto hasta la fecha)

The core Immune Profiler machine learning framework which will be at the heart of the MEDIVAC software solution is already beyond the current state-of-the-art in terms of its predictive capability. The adaption of this framework to develop the MEDIVAC software promises to help revolutionize cancer treatment allowing the development of personalized cancer vaccines in a cost effective and clinically actionable timeframe.

medivac-software-image.jpg