Predicting protein model correctness in Coot using machine learning

Abstract

Manually identifying and correcting errors in protein models can be a slow process, but improvements in validation tools and automated model-building software can contribute to reducing this burden. This article presents a new correctness score that is produced by combining multiple sources of information using a neural network. The residues in 639 automatically built models were marked as correct or incorrect by comparing them with the coordinates deposited in the PDB. A number of features were also calculated for each residue using Coot, including map-to-model correlation, density values, B factors, clashes, Ramachandran scores, rotamer scores and resolution. Two neural networks were created using these features as inputs: one to predict the correctness of main-chain atoms and the other for side chains. The 639 structures were split into 511 that were used to train the neural networks and 128 that were used to test performance. The predicted correctness scores could correctly categorize 92.3% of the main-chain atoms and 87.6% of the side chains. A Coot ML Correctness script was written to display the scores in a graphical user interface as well as for the automatic pruning of chains, residues and side chains with low scores. The automatic pruning function was added to the CCP4i2 Buccaneer automated model-building pipeline, leading to significant improvements, especially for high-resolution structures.

Allosteric modulation of cardiac myosin mechanics and kinetics by the conjugated omega‐7,9 trans‐fat rumenic acid

“…2010; Bond et al . 2020) for analysis, and UCSF Chimera 1.14 for visualization (Pettersen et al . 2004).…”

“…The resulting ensembles of top rumenic acid binding poses were selected according to the low-binding energy cut-off, equal to one hydrogen bond (≤−3.0 kcal mol −1 ). These ensembles were exported to PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC), COOT (Emsley et al 2010;Bond et al 2020) for analysis, and UCSF Chimera 1.14 for visualization (Pettersen et al 2004).…”

Residue Assignment in Crystallographic Protein Electron Density Maps With 3D Convolutional Networks

“…Earlier model building methods using machine learning relied on pattern matching [14] or feature-based approaches [15] [16]. In [17], neural networks are used to validate models built using Coot [18]. Protein electron density maps are usually represented [7] as three-dimensional volumetric data which is common in bio-imaging (e.g.…”

Predicting the performance of automated crystallographic model-building pipelines

“…We used data sets from three sources to train and evaluate our ML predictive model: 1203 experimental phasing data sets from the Joint Center for Structural Genomics (JCSG; van den Bedem et al, 2011;Alharbi et al, 2019), 32 newer experimental phasing data sets deposited between 2015 and 2021 and taken from the PDB, and 1332 molecular-replacement (MR) data sets from Bond et al (2020). These data sets correspond to two techniques that can be used to build a protein structure.…”

“…(v) Sequence identity: the sequence identity calculated by superposition of the homologue chain onto the target chain using GESAMT (Krissinel, 2012;Bond et al, 2020).…”