Alsaggaf, I. | Search bioRxiv

bioRxiv2024

Functional yeast promoter sequence design using temporal convolutional generative language models

Functional promoter sequence design plays a crucial role in accurately controlling gene expression processes that are one of the most fundamental mechanisms in biological systems. Thanks to the recent community effort, we are now able to elucidate the associations between yeast promoter sequences and their corresponding expression levels using advanced deep learning methods. This milestone boosts the further development of many downstream biological sequence research tasks including synthetic DNA sequence design. In this work, we propose a novel synthetic promoter sequence generation method, namely Gen-DNA-TCN, which exploits a pre-trained sequence-to-expression predictive model to facilitate its autoregressive generative model training. A large-scale evaluation confirms that Gen-DNA-TCN successfully generates a large number of unique, diverse and functional synthetic yeast promoter sequences that also encode similar transcription factor binding site distributions compared with real yeast promoter sequences.

bioinformatics↗

bioRxiv2022

Improving cell-type identification with Gaussian noise-augmented single-cell RNA-seq contrastive learning

Cell-type identification is an important task for single-cell RNA-seq (scRNA-seq) data analysis. In this work, we proposed a novel Gaussian noise augmented scRNA-seq contrastive learning framework (GsRCL) to learn a type of discriminative feature representations for cell-type prediction tasks. The experimental results suggest that the feature representations learned by GsRCL successfully improved the accuracy of cell-type prediction using scRNA-seq expression profiles.

bioinformatics↗