GAMA: A Robust and Automated Fragment-Based Quantum Chemistry Method for Biomolecular Systems

Accurate quantum chemical treatment of covalently bonded biomolecules using fragment-based approaches remains a major challenge as fragmenting across covalent bonds disrupts essential electron correlation and long-range polarization. Importantly, a few of the previously developed fragment-based methods can accurately and efficiently treat both noncovalently and covalently bonded molecular systems, highlighting a significant gap in the field. A key novelty of the grid-adapted many-body analysis (GAMA) framework is that it overcomes this limitation. Building on our earlier work establishing GAMA for noncovalent systems, we extend this framework to covalently bonded biomolecules and develop GAMA2, a fully automated protocol that integrates a simple grid-based fragmentation scheme, many-body expansion with overlapping fragments truncated at two-body order, and a multilayer low-level correction. Across diverse peptides, ranging from flexible bioactive motifs to structured 18-mer helixes, GAMA2 reproduces supersystem MP2/6-311G(d,p) energies with unsigned absolute errors of ∼0.01-4 kcal/mol for flexible small- and medium-size peptide systems using HF as a low level of theory and ∼2-5 kcal/mol for complicated helical-type peptide structures when using M06-2X/6-311G(d,p) as the low-level method, showing substantial improvement over HF using accurate DFT-based methods. In addition to this highly accurate results, GAMA2 also demonstrate a significant computational speedup with HF as a super system low-level method relative to the reference full MP2 calculation, establishing GAMA2 as a scalable, efficient, and systematically improvable route for correlated quantum chemical calculations on biomolecular systems.