Fig. 2
From: AttCRISPR: a spacetime interpretable model for prediction of sgRNA on-target activity
The architecture of spatial domain method in AttCRISPR. The input of the method is an encoded sgRNA sequence Xoh, a 21 × 4 one-hot matrix. Then refine it through a spatial attention module, which could tell us the importance of a specific matrix element (or just say, pixel). A simple CNN followed is applied to extract potential feature representation of sgRNA sequence. In the last step, we flatten the output of the CNN into a one-dimensional vector and use a multilayer perceptron with a sigmoid activation function to achieve the spatial output ys