Exploiting Graphoelements and Convolutional Neural Networks with Long Short Term Memory for Classification of the Human Electroencephalogram.

The electroencephalogram (EEG) is a cornerstone of neurophysiological research and clinical neurology. Historically, the classification of EEG as showing normal physiological or abnormal pathological activity has been performed by expert visual review. The potential value of unbiased, automated EEG classification has long been recognized, and in recent years the application of machine learning methods has received significant attention. A variety of solutions using convolutional neural networks (CNN) for EEG classification have emerged with impressive results. However, interpretation of CNN results and their connection with underlying basic electrophysiology has been unclear. This paper proposes a CNN architecture, which enables interpretation of intracranial EEG (iEEG) transients driving classification of brain activity as normal, pathological or artifactual. The goal is accomplished using CNN with long short-term memory (LSTM). We show that the method allows the visualization of iEEG graphoelements with the highest contribution to the final classification result using a classification heatmap and thus enables review of the raw iEEG data and interpret the decision of the model by electrophysiology means.

Nucleosome transactions on the Hypocrea jecorina (Trichoderma reesei) cellulase promoter cbh2 associated with cellulase induction.

The 5' regulatory region of the cbh2 gene of Hypocrea jecorina contains the cbh2 activating element (CAE) which is essential for induction of cbh2 gene expression by sophorose and cellulose. The CAE consists of two motifs, a CCAAT box on the template strand and a GTAATA box on the coding strand, which cooperate during induction. Northern analyses of cbh2 gene expression has revealed an absolute dependence on induction, but no direct effect of Cre1-mediated carbon catabolite repression. Investigation of the chromatin structure in the wild-type strain showed that, under repressing conditions, there is a nucleosome free region (nfr) around the CAE, which is flanked by strictly positioned nucleosomes. Induction results in a loss of positioning of nucleosomes -1 and -2 downstream of the CAE, thus making the TATA box accessible. Simultaneous mutation of both motifs of the CAE, or of the CCAAT-box alone, also leads to shifting of nucleosome -1, which normally covers the TATA-box under repressing conditions, whereas mutation of the GTAATA element results in a narrowing of the nfr, indicating that the proteins that bind to both motifs in the CAE interact with chromatin, although in different ways. A cellulase-negative mutant strain, which has previously been shown to be altered in protein binding to the CAE, still displayed the induction-specific changes in nucleosome structure, indicating that none of the proteins that directly interact with CAE are affected, and that nucleosome rearrangement and induction of cbh2 expression are uncoupled. Interestingly, the carbon catabolite repressor Cre1 is essential for strict nucleosome positioning in the 5' regulatory sequences of cbh2 under all of the conditions tested, and induction can occur in a promoter that lacks positioned nucleosomes. These data suggest that Cre1, the Hap2/3/5 complex and the GTAATA-binding protein are all involved in nucleosome assembly on the cbh2 promoter, and that the latter two respond to inducing conditions by repositioning nucleosome -1.