Distance‐based reconstruction of protein quaternary structures from inter‐chain contacts

Soltanikazemi, Elham; Quadir, Farhan; Roy, Raj; Guo, Zhiye; Cheng, Jianlin

doi:10.1002/prot.26269

Cited by 11 publications

(7 citation statements)

References 63 publications

(64 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the residual blocks, a 2D convolutional layer with the softmax function is used to classify the distance between any two residues from two monomers in a dimer into 42 distance bins (i.e., 40 bins from 2 to 22 Å with a bin size of 0.5 Å, plus a 0–2 Å bin and a >22 Å bin). Two kinds of inter-chain residue-residue distance are predicted at the same time: (1) the distance between the two closest heavy atoms from two residues used by most existing works in the field and (2) the C b -C b distance between two residues used by some recent works 36 , resulting in two kinds of distance maps predicted.…”

Section: Methodsmentioning

confidence: 99%

Prediction of inter-chain distance maps of protein complexes with 2D attention-based deep neural networks

et al. 2022

Self Cite

View full text Add to dashboard Cite

Residue-residue distance information is useful for predicting tertiary structures of protein monomers or quaternary structures of protein complexes. Many deep learning methods have been developed to predict intra-chain residue-residue distances of monomers accurately, but few methods can accurately predict inter-chain residue-residue distances of complexes. We develop a deep learning method CDPred (i.e., Complex Distance Prediction) based on the 2D attention-powered residual network to address the gap. Tested on two homodimer datasets, CDPred achieves the precision of 60.94% and 42.93% for top L/5 inter-chain contact predictions (L: length of the monomer in homodimer), respectively, substantially higher than DeepHomo’s 37.40% and 23.08% and GLINTER’s 48.09% and 36.74%. Tested on the two heterodimer datasets, the top Ls/5 inter-chain contact prediction precision (Ls: length of the shorter monomer in heterodimer) of CDPred is 47.59% and 22.87% respectively, surpassing GLINTER’s 23.24% and 13.49%. Moreover, the prediction of CDPred is complementary with that of AlphaFold2-multimer.

show abstract

Section: Methodsmentioning

confidence: 99%

Prediction of inter-chain distance maps of protein complexes with 2D attention-based deep neural networks

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…We compare DRLComplex with four existing methods namely GD (gradient descent) (Soltanikazemi et al 2022), MC (Markov Chain Monte Carlo simulation), Equidock (Ganea et al 2022) and CNS (Brünger et al 1998;Brunger 2007) (the simulated annealing simulation used in ConComplex) (Quadir, Roy, Soltanikazemi, et al 2021) that constructs quaternary structures. It is worth noting that Equidock is an equivariant neural network method of directly predicting quaternary structures of dimers without using interchain contacts as input.…”

Section: Resultsmentioning

confidence: 99%

“…The methods are evaluated on two standard protein dimer datasets: CASP_CAPRI and Std_32. The CASP_CAPRI dataset consists of 28 homodimer(Yan and Huang 2021) targets and the Std_32 contains 32 heterodimer targets (Soltanikazemi et al 2022). It is worth noting that only 31 out of the 32 targets of the Std_32 are used because one target (1IXRA_1IXRC) has no interchain contact between its ligand and receptor.…”

Section: Resultsmentioning

confidence: 99%

“…Zdock (Pierce et al 2014) uses a combination of various potential scoring criteria to select the quaternary structure models generated by the Fourier transformation. Recently, a gradient descent structure optimization method -GD (Soltanikazemi et al 2022) and a simulated annealing method -ConComplex (Quadir, Roy, Soltanikazemi, et al 2021) have been developed to build quaternary structures from predicted interchain residue-residue contacts and tertiary structures of protein chains. Due to the recent development of more accurate deep learning methods for interchain contact prediction such as DRCon (Roy et al 2021) and GLINTER (Xie and Xu 2021), these approaches can generate high-quality quaternary structures for some proteins.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A deep reinforcement learning approach to reconstructing quaternary structures of protein dimers through self-learning

Soltanikazemi

Roy

Quadir

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Predicted interchain residue-residue contacts can be used to build the quaternary structure of protein complexes from scratch. However, only a small number of methods have been developed to reconstruct protein quaternary structures using predicted interchain contacts. Here, we present an agent-based self-learning method based on deep reinforcement learning (DRLComplex) to build protein complex structures using interchain contacts as distance constraints. We rigorously tested the DRLComplex on two standard datasets of homodimeric and heterodimeric dimers (the CASP-CAPRI homodimer dataset and Std_32 heterodimer dataset) using both true and predicted contacts. Utilizing true contacts as input, the DRLComplex achieved a high average TM-score of 0.9895 and 0.9881 and a low average interface RMSD (I_RMSD) of 0.2197 and 0.92 on the two datasets, respectively. When predicted contacts are used, the method achieves the TM-score of 0.73 and 0.76 for homodimers and heterodimers respectively. The accuracy of reconstructed quaternary structures depends on the accuracy of contact predictions. Compared with other optimization methods of reconstructing quaternary structures from interchain contacts, DRLComplex performs similarly to an advanced gradient descent method and better than a Markov Chain Monte Carlo simulation method and a simulated annealing-based method. The source code of DRLComplex is available at: https://github.com/jianlin-cheng/DRLComplex

show abstract

“…For the human complex V subunit MT-ATP6, the 3D structure predicted by AlphaFold obtained from UniProt was used (AF-PYE00846-F1) 69 . A protein residue was regarded to be in close proximity to regional constraint when the minimum distance between its alpha carbon atom and the alpha carbon of a residue in regional constraint was <6 Ångstrom, a threshold commonly used to define contacting residues 70 . A rRNA base was regarded to be in close proximity to regional constraint when the minimum distance between its nitrogen atom involved in base pairing and the equivalent nitrogen of a base in regional constraint was <6 Ångstrom, or if its phosphate atom and the phosphate of a base in regional constraint was <6 Ångstrom.…”

Section: Methodsmentioning

confidence: 99%

Quantifying constraint in the human mitochondrial genome

Lake

Liu

Battle

et al. 2022

Preprint

View full text Add to dashboard Cite

Mitochondrial DNA (mtDNA) has an important, yet often overlooked, role in health and disease. Constraint models quantify depletion of genetic variation by negative selection, representing a powerful tool for identifying deleterious variation underlying human phenotypes1-4. However, a constraint model for the mtDNA had not been developed, due to the unique features of this genome. Here we describe the development of a mitochondrial constraint model and its application to the Genome Aggregation Database (gnomAD), a large-scale population dataset reporting mtDNA variation across 56,434 humans5. Our results demonstrate strong depletion of expected variation, showing most pathogenic mtDNA variants remain undiscovered. To aid their identification, we compute constraint metrics for every protein, transfer RNA, and ribosomal RNA gene in the human mtDNA, as well as for non-coding regions. This includes assessment of regional or positional constraint within every gene, and local constraint for every base position. We identify enrichment of pathogenic variation in the most constrained sites, which include loci typically overlooked in mtDNA analyses, and show that the most constrained gene regions and non-coding elements encode functionally-critical sites. Lastly, we demonstrate how these metrics can improve the discovery of mtDNA variation underlying rare and common human phenotypes.

show abstract

Distance‐based reconstruction of protein quaternary structures from inter‐chain contacts

Cited by 11 publications

References 63 publications

Prediction of inter-chain distance maps of protein complexes with 2D attention-based deep neural networks

Prediction of inter-chain distance maps of protein complexes with 2D attention-based deep neural networks

A deep reinforcement learning approach to reconstructing quaternary structures of protein dimers through self-learning

Quantifying constraint in the human mitochondrial genome

Contact Info

Product

Resources

About