Sequence basis of transcription initiation in the human genome.

Transcription initiation is a process that is essential to ensuring the proper function of any gene, yet we still lack a unified understanding of sequence patterns and rules that explain most transcription start sites in the human genome. By predicting transcription initiation at base-pair resolution from sequences with a deep learning-inspired explainable model called Puffin, we show that a small set of simple rules can explain transcription initiation at most human promoters. We identify key sequence patterns that contribute to human promoter activity, each activating transcription with distinct position-specific effects. Furthermore, we explain the sequence basis of bidirectional transcription at promoters, identify the links between promoter sequence and gene expression variation across cell types, and explore the conservation of sequence determinants of transcription initiation across mammalian species.

Sequence basis of transcription initiation in human genome.

Transcription initiation is an essential process for ensuring proper function of any gene, however, a unified understanding of sequence patterns and rules that determine transcription initiation sites in human genome remains elusive. By explaining transcription initiation at basepair resolution from sequence with a deep learning-inspired explainable modeling approach, here we show that simple rules can explain the vast majority of human promoters. We identified key sequence patterns that contribute to human promoter function, each activating transcription with a distinct position-specific effect curve that likely reflects its mechanism of promoting transcription initiation. Most of these position-specific effects have not been previously characterized, and we verified them using experimental perturbations of transcription factors and sequences. We revealed the sequence basis of bidirectional transcription at promoters and links between promoter selectivity and gene expression variation across cell types. Additionally, by analyzing 241 mammalian genomes and mouse transcription initiation site data, we showed that the sequence determinants are conserved across mammalian species. Taken together, we provide a unified model of the sequence basis of transcription initiation at the basepair level that is broadly applicable across mammalian species, and shed new light on basic questions related to promoter sequence and function.

Dirichlet Diffusion Score Model for Biological Sequence Generation.

Designing biological sequences is an important challenge that requires satisfying complex constraints and thus is a natural problem to address with deep generative modeling. Diffusion generative models have achieved considerable success in many applications. Score-based generative stochastic differential equations (SDE) model is a continuous-time diffusion model framework that enjoys many benefits, but the originally proposed SDEs are not naturally designed for modeling discrete data. To develop generative SDE models for discrete data such as biological sequences, here we introduce a diffusion process defined in the probability simplex space with stationary distribution being the Dirichlet distribution. This makes diffusion in continuous space natural for modeling discrete data. We refer to this approach as Dirchlet diffusion score model. We demonstrate that this technique can generate samples that satisfy hard constraints using a Sudoku generation task. This generative model can also solve Sudoku, including hard puzzles, without additional training. Finally, we applied this approach to develop the first human promoter DNA sequence design model and showed that designed sequences share similar properties with natural promoter sequences.