Polygraph: A Software Framework for the Systematic Assessment of Synthetic Regulatory DNA Elements

Abstract

The design of regulatory elements is pivotal in numerous therapeutic interventions, including gene and cell therapy, wherein the typical objective is to engineer DNA sequences exhibiting specific attributes like cell-type specificity and elevated expression levels. However, the systematic assessment of these constructed DNA sequences remains challenging due to the absence of robust metrics and an integrated software framework. Here, we introduce Polygraph, a Python framework for evaluating synthetic DNA sequences. Polygraph provides a variety of features to streamline the synthesis and scrutiny of regulatory elements, incorporating features like a diversity index, motif and k-mer composition, similarity to endogenous regulatory sequences, and screening with predictive and foundational models. Consequently, Polygraph stands as the first instrument for the assessment of synthetic regulatory sequences, enabling expedited progress in therapeutic applications and improving our understanding of gene regulatory mechanisms.

regLM: Designing realistic regulatory DNA with autoregressive language models

“…For a fair comparison, we performed all five model-guided methods using the regression model trained on the same dataset as regLM as an oracle. All sets of synthetic promoters were compared to known strong promoters from the test set using Polygraph [11]. Below, we use Evolution (V) to refer to synthetic promoters generated by Vaishnav et al [20] using Directed Evolution.…”