# Probing and Perturbing Transcriptional Condensates with Multiscale Modeling and Deep Learning

> **NIH NIH R35** · MASSACHUSETTS INSTITUTE OF TECHNOLOGY · 2024 · $405,713

## Abstract

PROJECT SUMMARY
Biomolecular condensates are compartments that form within cells through the self-assembly of proteins, RNAs,
etc. These condensates play a critical role in the functioning of cells as they help to concentrate biomolecules
and facilitate biochemical reactions. Recent studies have shown that they are also essential for transcriptional
regulation. However, there are still many unanswered questions about how these condensates form, interact
with each other, and how they organize chromatin and regulate transcription. One of the challenges in addressing
these questions is the lack of suitable experimental techniques for quantitatively studying the spatiotemporal
complexity and heterogeneity of transcriptional condensates. We propose to characterize transcriptional
condensates in the context of chromatin using in silico approaches. (i) By introducing many-body potentials
represented using graph neural networks and novel parameterization schemes, we will introduce the next
generation of residue-level coarse-grained (CG) models for proteins, RNAs, and DNAs. These models will
outperform existing ones that often lack the accuracy needed for de novo predictions, enabling their application
from generating testable hypotheses. (ii) Using CG models that balance computational efficiency and chemical
accuracy, we will reconstruct in vivo chromatin organization at near-atomistic resolution with a multiscale
approach. We will evaluate the stability of these near-atomistic models and the contribution of physicochemical
interactions in stabilizing chromatin. In addition, we will quantify the role of transcriptional condensates in
chromatin looping and accelerating transcription factor target search using reconstructed in vivo chromatin
structures. (iii) We will also decode the molecular grammar of protein-protein interactions that dictate condensate
stability using evolutionary sequence analysis and physical simulations. We will use this information to design
small molecules for condensate modulation. Overall, the goal of our research is to provide critical evaluations of
hypotheses on the role of transient, collective phenomena arising from weak, multivalent interactions in essential
cellular processes and suggest novel therapeutic strategies for their modulation.

## Key facts

- **NIH application ID:** 10842909
- **Project number:** 2R35GM133580-06
- **Recipient organization:** MASSACHUSETTS INSTITUTE OF TECHNOLOGY
- **Principal Investigator:** Bin Zhang
- **Activity code:** R35 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $405,713
- **Award type:** 2
- **Project period:** 2019-08-01 → 2029-05-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10842909

## Citation

> US National Institutes of Health, RePORTER application 10842909, Probing and Perturbing Transcriptional Condensates with Multiscale Modeling and Deep Learning (2R35GM133580-06). Retrieved via AI Analytics 2026-05-26 from https://api.ai-analytics.org/grant/nih/10842909. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*