The protein structure prediction from its query amino acid sequence has been considered for many year a very hard Bioinformatics problem. Recently [1] there has been a breakthrough on this area thanks to the incorporation of "Direct Coupling Analysis" (DCA, based on coevolution) that allows to predict long-distance contacts and consequently to estimate a Contact Map (CMap) image. This image can be considered noisy and recently we have enhanced it using Deep Convolutional Neural Networks. In this talk we will discuss the possibility of using GANs to enhance this image.
Forschungs-