NLM DIR Seminar Schedule
UPCOMING SEMINARS
-
Jan. 20, 2026 Anastasia Gulyaeva
TBD -
Jan. 22, 2026 Mario Flores
AI Pipeline for Characterization of the Tumor Microenvironment -
Jan. 27, 2026 Zhaohui Liang
TBD -
Jan. 29, 2026 Mehdi Bagheri Hamaneh
FastSpel: A simple peptide spectrum predictor that achieves deep learning-level performance at a fraction of the computational cost -
Feb. 3, 2026 Matthew Diller
TBD
RECENT SEMINARS
-
Jan. 8, 2026 Won Gyu Kim
LitSense 2.0: AI-powered biomedical information retrieval with sentence and passage level knowledge discovery -
Dec. 16, 2025 Sarvesh Soni
ArchEHR-QA: A Dataset and Shared Task for Grounded Question Answering from Electronic Health Records -
Dec. 2, 2025 Qingqing Zhu
CT-Bench & CARE-CT: Building Reliable Multimodal AI for Lesion Analysis in Computed Tomography -
Nov. 25, 2025 Jing Wang
MIMIC-EXT-TE: Millions Clinical Temporal Event Time-Series Dataset -
Oct. 21, 2025 Yifan Yang
TBD
Scheduled Seminars on June 10, 2025
In-person: Building 38A/B2N14 NCBI Library or Meeting Link
Contact NLMDIRSeminarScheduling@mail.nih.gov with questions about this seminar.
Abstract:
Open-shell systems, characterized by the presence of unpaired electrons, play a central role in many biological processes, including electron transfer, oxygen activation, and enzymatic catalysis. Their unique electronic structures give rise to distinct interaction patterns not observed in closed-shell systems, providing critical insights into fundamental molecular mechanisms in biology. Understanding the forces and energy landscapes associated with open-shell interactions enables prediction of reactivity and supports the rational design of new compounds.
Despite advances in methods for modeling molecular interactions—from molecular docking to molecular mechanics and quantum mechanical approaches—accurate treatment of open-shell and excited-state interactions remains challenging. We propose a combined ∆SCF and coupled-cluster (CC) approach as a practical and efficient method for computing state-specific interaction energies, particularly when specific electronic configurations must be defined.
Moreover, given the computational cost of quantum methods, predicting the attractive or repulsive nature of pre-bonding interactions using classical electrostatics remains an attractive alternative, especially for large biological systems. Accuracy can be improved by incorporating quantum-derived properties, such as atomic polarizabilities, as seen in modern force fields. In this work, we extend our group’s Small Dielectric Spheres Model to open-shell systems with lone p-electrons and demonstrate its superior accuracy over common DFT methods at pre-bonding distances. We further argue that treating molecular systems as classical dielectrics offers a promising direction for modeling pre-bonding interactions in ligands, proteins, and cellular membranes.