Computational Protein Structures using MD, AI​/ML HDX-MS Intern

Position: Computational Protein Structures using MD, AI/ML with HDX-MS Intern
Location: Spring House

At Johnson & Johnson, we believe health is everything. Our strength in healthcare innovation empowers us to build a world where complex diseases are prevented, treated, and cured, where treatments are smarter and less invasive, and solutions are personal. Through our expertise in Innovative Medicine and Med Tech, we are uniquely positioned to innovate across the full spectrum of healthcare solutions today to deliver the breakthroughs of tomorrow, and profoundly impact health for humanity.

Learn more at

Job Function:
Career Programs
Job Sub Function:
Non-LDP Intern/Co-Op
Job Category:
Career Program
All Job Posting Locations:
Spring House, Pennsylvania, United States of America
Job Description:

About Innovative Medicine

Our expertise in Innovative Medicine is informed and inspired by patients, whose insights fuel our science-based advancements. Visionaries like you work on teams that save lives by developing the medicines of tomorrow.

Join us in developing treatments, finding cures, and pioneering the path from lab to life while championing patients every step of the way. Learn more at

We are searching for the best talent for on Computational Characterization of Protein Structures using MD and AI/ML with HDX-MS Intern to be in Spring House, PA.

• The Intern term is from June to August, 2026.
• Full time requirement (40 hours per week).

Purpose:

Accurate prediction of protein and protein complex structures is essential throughout structure-based drug discovery. Recent AI/ML advances significantly improved prediction accuracy and efficiency, particularly diffusion-based co-folding methods like Alpha

Fold3, Boltz-2, Chai-2, RoseTTAFold All-Atom, and Open Fold-3, now enable reliable in silico predictions of experimentally validated structures and key binding sites for diverse biomolecular complexes including protein-small molecule, protein-PROTAC/molecular glue ternary complexes, protein-peptide, protein-nucleic acid, and antibody-antigen complexes. These tools generate both static structures and conformational ensembles, facilitating integration with molecular dynamics simulations to capture system dynamics in understanding the mechanism of action (MOA).

Despite their promising utility, these predictions still require experimental confirmation (e.g., X-ray or Cryo-EM), which often relegates their application to retrospective studies and limits their impact in early-phase drug discovery.

Hydrogen-deuterium exchange mass spectrometry (HDX-MS) is increasingly used to study protein-protein and protein-ligand interactions by measuring the exchange rate of backbone amide hydrogens with deuterium, reflecting flexibility and solvent exposure. HDX-MS offers insights into protein structure and dynamics, complementing high-resolution X-ray data. It is easier and earlier to use than X-ray or Cryo-EM in early drug discovery. Recently, HDX-MS has seen broader application in physics-based simulations, and AI/ML models now reconstruct HDX data and help prioritize protein structure models by matching experimental results to structures.

Co-folding models and HDX-MS are widely used at Johnson & Johnson, driving both large and small molecule modalities. Multiple co-folding platforms have been established, and extensive HDX-MS spectra for various protein targets, along with physics-based computational analysis tools, have been developed. By further integrating AI/ML capabilities and evaluating physics- and ML-based HDX methods for prioritizing co-folding models, structure-based design can be improved before X-ray or Cryo-EM data are available.

This approach will accelerate our drug discovery cycle and enhances collaboration with internal and external partners.

The main objective of this project aims to integrate public HDX modeling methods with internal physics-based approaches in the JNJ platform to improve selection of correct bound complex structures, reweight MOA-related conformational ensembles, and identify potential cryptic pockets for allosteric protein-ligand binding. The student will create benchmark datasets of co-folding models, develop Python scripts to automate both public and internal HDX-MS modeling methods, and evaluate them through prioritizing co-folding models in modality-agnostic targets, especially focus on antibodies and antigens.

The student is expected to work in a team environment, having direct interaction and good communication with people from different departments within JJIM.

The student will gain experience with physics- and AI/ML-based methods, understand HDX-MS data, and learn how it informs protein structure selection. The enhanced platform will streamline co-folding model prioritization and increase effective use of co-folding models and HDX-MS data in drug discovery projects at JJIM.

You will be responsible

• Develop Python scripts to automate the integration of public methods into the JNJ computational platform.
• Perform scientific computations, such as co-folding and molecular dynamics, to generate protein structure…