Robustly interrogating machine learning-based scoring functions: what are they learning?

Motivation: Machine learning-based scoring functions (MLBSFs) have been found to exhibit inconsistent performance on different benchmarks and be prone to learning dataset bias. For the field to develop MLBSFs that learn a generalisable understanding of physics, a more rigorous understanding of how they perform is required.

Results: In this work, we compared the performance of a diverse set of popular MLBSFs (RFScore, SIGN, OnionNet-2, Pafnucy, and PointVS) to our proposed baseline models that can only learn dataset biases on a range of benchmarks.

Challenges and compromises: Predicting unbound antibody structures with deep learning.

Therapeutic antibodies are manufactured, stored and administered in the free state; this makes understanding the unbound form key to designing and improving development pipelines. Prediction of unbound antibodies is challenging, specifically modelling of the CDRH3 loop, where inaccuracies are potentially worse due to a bias in structural data towards antibody-antigen complexes. This class imbalance provides a challenge for deep learning models trained on this data, potentially limiting generalisation to unbound forms.

Fragmenstein: predicting protein-ligand structures of compounds derived from known crystallographic fragment hits using a strict conserved-binding-based methodology.

Current strategies centred on either merging or linking initial hits from fragment-based drug design (FBDD) crystallographic screens generally do not fully leaverage 3D structural information. We show that an algorithmic approach (Fragmenstein) that 'stitches' the ligand atoms from this structural information together can provide more accurate and reliable predictions for protein-ligand complex conformation than general methods such as pharmacophore-constrained docking. This approach works under the assumption of conserved binding: when a larger molecule is designed containing the initial fragment hit, the common substructure between the two will adopt the same binding mode.

The molecular reach of antibodies crucially underpins their viral neutralisation capacity.

Key functions of antibodies, such as viral neutralisation, depend on high-affinity binding. However, viral neutralisation poorly correlates with antigen affinity for reasons that have been unclear. Here, we use a new mechanistic model of bivalent binding to study  >45 patient-isolated IgG1 antibodies interacting with SARS-CoV-2 RBD surfaces.

Antibody Design with SE(3) Diffusion.

We introduce , an antibody variable domain diffusion model based on a general protein backbone diffusion framework, which was extended to handle multiple chains. Assessing the designability and novelty of the structures generated with our model, we find that produces highly designable antibodies that can contain novel binding regions. The backbone dihedral angles of sampled structures show good agreement with a reference antibody distribution.