Fig. 1
From: Machine learning predicts nucleosome binding modes of transcription factors
A model to predict nucleosome-TF binding patterns. The sequence \(s_{i}\) is projected to a multivariate Gaussian distribution with mean \(\tilde{\mu }_{i}\) and covariance matrix \(\tilde{\Sigma }_{i}\). Similarity between these multivariate Gaussian distributions is computed to form kernel matrix and a multi-label classifier is trained to model binding preferences using the kernel matrix