Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Co-evolution-based prediction of metal-binding sites in proteomes by machine learning

Nature Chemical Biology (2023)Cite this article

Subjects

Abstract

Metal ions have various important biological roles in proteins, including structural maintenance, molecular recognition and catalysis. Previous methods of predicting metal-binding sites in proteomes were based on either sequence or structural motifs. Here we developed a co-evolution-based pipeline named ‘MetalNetʼ to systematically predict metal-binding sites in proteomes. We applied MetalNet to proteomes of four representative prokaryotic species and predicted 4,849 potential metalloproteins, which substantially expands the currently annotated metalloproteomes. We biochemically and structurally validated previously unannotated metal-binding sites in several proteins, including apo-citrate lyase phosphoribosyl-dephospho-CoA transferase citX, an Escherichia coli enzyme lacking structural or sequence homology to any known metalloprotein (Protein Data Bank (PDB) codes: 7DCM and 7DCN). MetalNet also successfully recapitulated all known zinc-binding sites from the human spliceosome complex. The pipeline of MetalNet provides a unique and enabling tool for interrogating the hidden metalloproteome and studying metal biology.

This is a preview of subscription content, access via your institution

Access options

Rent or buy this article

Get just this article for as long as you need it

$39.95

Learn more

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Predicting metal-binding residue pairs from co-evolution by ML.
Fig. 2: Assembly of metal-binding residue pairs into a high-order connected co-evolution network by graph theory.
Fig. 3: Predictions and analyses of metalloproteins from representative prokaryotic species.
Fig. 4: Biochemical and structural characterization of citX as a zinc-binding protein.

Data availability

The original protein structure dataset containing 9,846 protein sequences and their MSA can be downloaded from https://doi.org/10.1073/pnas.1702664114. Co-evolved pairs used in model training can be found in Supplementary Dataset. The proteins and related MSA of prokaryotic species can be downloaded from https://gremlin2.bakerlab.org/db/{species}/fasta/. The Metagenome-pfam MSA and structural model can be downloaded from https://gremlin2.bakerlab.org/db/UNI/. The human spliceosome dataset can be found in Supplementary Dataset. PDB codes 6ID0, 6ID1, 6ICZ and 6QW6 were used to construct human spliceosome dataset. We downloaded the information table of protein entities from the PDB server (https://ftp.wwpdb.org/pub/pdb/derived_data/index/entries.idx) to construct unbiased dataset when comparing methods. UniProt (https://ebi10.uniprot.org) profiles (date: 16 August 2021), Gene Ontology database (https://www.ebi.ac.uk/QuickGO/) and Pfam database (http://pfam.xfam.org/) were used in the analysis. The pdbaa database (16 January 2018 release, ftp://ftp.ncbi.nlm.nih.gov/blast/db/pdbaa.tar.gz) is used in BLASTP. The structures of citX reported in this paper have been deposited in PDB with the accession numbers 7DCM (determined by single-wavelength anomalous diffraction) and 7DCN (determined by molecular replacement). Source data are provided with this paper.

Code availability

Our code is available as open source at https://github.com/wangchulab/MetalNet.

References

  1. Gladyshev, V. N. & Zhang, Y. Comparative genomics analysis of the metallomes. Met. Ions Life Sci. 12, 529–580 (2013).

    Article  Google Scholar 

  2. Waldron, K. J. & Robinson, N. J. How do bacterial cells ensure that metalloproteins get the correct metal? Nat. Rev. Microbiol. 7, 25–35 (2009).

    Article  CAS  Google Scholar 

  3. Yannone, S. M., Hartung, S., Menon, A. L., Adams, M. W. & Tainer, J. A. Metals in biology: defining metalloproteomes. Curr. Opin. Biotechnol. 23, 89–95 (2012).

    Article  CAS  Google Scholar 

  4. Waldron, K. J., Rutherford, J. C., Ford, D. & Robinson, N. J. Metalloproteins and metal sensing. Nature 460, 823–830 (2009).

    Article  CAS  Google Scholar 

  5. Cvetkovic, A. et al. Microbial metalloproteomes are largely uncharacterized. Nature 466, 779–782 (2010).

    Article  CAS  Google Scholar 

  6. Pace, N. J. & Weerapana, E. A competitive chemical-proteomic platform to identify zinc-binding cysteines. ACS Chem. Biol. 9, 258–265 (2014).

    Article  CAS  Google Scholar 

  7. Sevcenco, A. M. et al. Exploring the microbial metalloproteome using MIRAGE. Metallomics 3, 1324–1330 (2011).

    Article  CAS  Google Scholar 

  8. Andreini, C., Banci, L., Bertini, I. & Rosato, A. Counting the zinc-proteins encoded in the human genome. J. Proteome Res. 5, 196–201 (2006).

    Article  CAS  Google Scholar 

  9. Passerini, A., Punta, M., Ceroni, A., Rost, B. & Frasconi, P. Identifying cysteines and histidines in transition‐metal‐binding sites using support vector machines and neural networks. Proteins Struct. Funct. Bioinf. 65, 305–316 (2006).

    Article  CAS  Google Scholar 

  10. Passerini, A., Lippi, M. & Frasconi, P. MetalDetector v2.0: predicting the geometry of metal binding sites from protein sequence. Nucleic Acids Res. 39, W288–W292 (2011).

    Article  CAS  Google Scholar 

  11. Haberal, İ. & Oğul, H. Prediction of protein metal binding sites using deep neural networks. Mol. Inf. 38, e1800169 (2019).

    Article  Google Scholar 

  12. Babor, M., Gerzon, S., Raveh, B., Sobolev, V. & Edelman, M. Prediction of transition metal-binding sites from apoprotein structures. Proteins 70, 208–217 (2008).

    Article  CAS  Google Scholar 

  13. Lin, Y. F. et al. MIB: metal ion-binding site prediction and docking server. J. Chem. Inf. Model. 56, 2287–2291 (2016).

    Article  CAS  Google Scholar 

  14. Zhang, C., Freddolino, P. L. & Zhang, Y. COFACTOR: improved protein function prediction by combining structure, sequence and protein-protein interaction information. Nucleic Acids Res. 45, W291–W299 (2017).

    Article  CAS  Google Scholar 

  15. Gobel, U., Sander, C., Schneider, R. & Valencia, A. Correlated mutations and residue contacts in proteins. Proteins 18, 309–317 (1994).

    Article  CAS  Google Scholar 

  16. Shindyalov, I. N., Kolchanov, N. A. & Sander, C. Can three-dimensional contacts in protein structures be predicted by analysis of correlated mutations? Protein Eng. Des. Select. 7, 349–358 (1994).

    Article  CAS  Google Scholar 

  17. Martin, L. C., Gloor, G. B., Dunn, S. D. & Wahl, L. M. Using information theory to search for co-evolving residues in proteins. Bioinformatics 21, 4116–4124 (2005).

    Article  CAS  Google Scholar 

  18. Morcos, F. et al. Direct-coupling analysis of residue coevolution captures native contacts across many protein families. Proc. Natl Acad. Sci. USA 108, E1293–E1301 (2011).

    Article  CAS  Google Scholar 

  19. Balakrishnan, S., Kamisetty, H., Carbonell, J. G., Lee, S. I. & Langmead, C. J. Learning generative models for protein fold families. Proteins 79, 1061–1078 (2011).

    Article  CAS  Google Scholar 

  20. Jones, D. T., Buchan, D. W., Cozzetto, D. & Pontil, M. PSICOV: precise structural contact prediction using sparse inverse covariance estimation on large multiple sequence alignments. Bioinformatics 28, 184–190 (2012).

    Article  CAS  Google Scholar 

  21. Marks, D. S., Hopf, T. A. & Sander, C. Protein structure prediction from sequence variation. Nat. Biotechnol. 30, 1072–1080 (2012).

    Article  CAS  Google Scholar 

  22. Ovchinnikov, S. et al. Protein structure determination using metagenome sequence data. Science 355, 294–298 (2017).

    Article  CAS  Google Scholar 

  23. Xu, J. Distance-based protein folding powered by deep learning. Proc. Natl Acad. Sci. USA 116, 16856–16865 (2019).

    Article  CAS  Google Scholar 

  24. Ovchinnikov, S., Kamisetty, H. & Baker, D. Robust and accurate prediction of residue-residue interactions across protein interfaces using evolutionary information. eLife 3, e02030 (2014).

    Article  Google Scholar 

  25. Cong, Q., Anishchenko, I., Ovchinnikov, S. & Baker, D. Protein interaction networks revealed by proteome coevolution. Science 365, 185–189 (2019).

    Article  CAS  Google Scholar 

  26. Toth-Petroczy, A. et al. Structured states of disordered proteins from genomic sequences. Cell 167, 158–170 (2016).

    Article  CAS  Google Scholar 

  27. Chakrabarti, S. & Panchenko, A. R. Coevolution in defining the functional specificity. Proteins 75, 231–240 (2009).

    Article  CAS  Google Scholar 

  28. Kuipers, R. K. et al. Correlated mutation analyses on super-family alignments reveal functionally important residues. Proteins 76, 608–616 (2009).

    Article  CAS  Google Scholar 

  29. Chakrabarti, S. & Panchenko, A. R. Structural and functional roles of coevolved sites in proteins. PLoS One 5, e8591 (2010).

    Article  Google Scholar 

  30. Jeong, C. S. & Kim, D. Structure-based Markov random field model for representing evolutionary constraints on functional sites. BMC Bioinf. 17, 99 (2016).

    Article  Google Scholar 

  31. Wang, G. & Dunbrack, R. L. Jr. PISCES: a protein sequence culling server. Bioinformatics 19, 1589–1591 (2003).

    Article  CAS  Google Scholar 

  32. Anishchenko, I., Ovchinnikov, S., Kamisetty, H. & Baker, D. Origins of coevolution between residues distant in protein 3D structures. Proc. Natl Acad. Sci. USA 114, 9122–9127 (2017).

    Article  CAS  Google Scholar 

  33. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local alignment search tool. J. Mol. Biol. 215, 403–410 (1990).

    Article  CAS  Google Scholar 

  34. Cariss, S. J. L. et al. YieJ (CbrC) mediates CreBC-dependent colicin E2 tolerance in Escherichia coli. J. Bacteriol. 192, 3329–3336 (2010).

    Article  CAS  Google Scholar 

  35. Schneider, K., Dimroth, P. & Bott, M. Biosynthesis of the prosthetic group of citrate lyase. Biochemistry 39, 9438–9450 (2000).

    Article  CAS  Google Scholar 

  36. Will, C. L. & Luhrmann, R. Spliceosome structure and function. Csh Perspect. Biol. 3, a003707 (2011).

    CAS  Google Scholar 

  37. Charenton, C., Wilkinson, M. E. & Nagai, K. Mechanism of 5′ splice site transfer for human spliceosome activation. Science 364, 362–367 (2019).

    Article  CAS  Google Scholar 

  38. Zhang, X. F. et al. Structures of the human spliceosomes before and after release of the ligated exon. Cell Res. 29, 274–285 (2019).

    Article  CAS  Google Scholar 

  39. Zhang, C. X., Zheng, W., Mortuza, S. M., Li, Y. & Zhang, Y. DeepMSA: constructing deep multiple sequence alignment to improve contact prediction and fold-recognition for distant-homology proteins. Bioinformatics 36, 2105–2112 (2020).

    Article  CAS  Google Scholar 

  40. Piazza, I. et al. A map of protein-metabolite interactions reveals principles of chemical communication. Cell 172, 358–372 (2018).

    Article  CAS  Google Scholar 

  41. Zhuang, S., Li, Q., Cai, L., Wang, C. & Lei, X. Chemoproteomic profiling of bile acid interacting proteins. ACS Cent. Sci. 3, 501–509 (2017).

    Article  CAS  Google Scholar 

  42. Horning, B. D. et al. Chemical proteomic profiling of human methyltransferases. J. Am. Chem. Soc. 138, 13335–13343 (2016).

    Article  CAS  Google Scholar 

  43. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

    Article  CAS  Google Scholar 

  44. Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).

    Article  CAS  Google Scholar 

  45. Cock, P. J. A. et al. Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25, 1422–1423 (2009).

    Article  CAS  Google Scholar 

  46. Steinegger, M. et al. HH-suite3 for fast remote homology detection and deep protein annotation. BMC Bioinform. 20, 473 (2019).

    Article  Google Scholar 

  47. Varoquaux, G., Vaught, T., & Millman, J. (eds.). Exploring network structure, dynamics, and function using networkX. In Proceedings of the 7th Python in Science Conference 11–15 (SciPy, 2008).

  48. Huang, Y., Niu, B. F., Gao, Y., Fu, L. M. & Li, W. Z. CD-HIT suite: a web server for clustering and comparing biological sequences. Bioinformatics 26, 680–682 (2010).

    Article  CAS  Google Scholar 

  49. Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).

    Article  CAS  Google Scholar 

  50. Hulsen, T., de Vlieg, J. & Alkema, W. BioVenn—a web application for the comparison and visualization of biological lists using area-proportional Venn diagrams. BMC Genom. 9, 488 (2008).

    Article  Google Scholar 

  51. Song, Y. F. et al. High-resolution comparative modeling with RosettaCM. Structure 21, 1735–1742 (2013).

    Article  CAS  Google Scholar 

  52. Wang, C., Vernon, R., Lange, O., Tyka, M. & Baker, D. Prediction of structures of zinc-binding proteins through explicit modeling of metal coordination geometry. Protein Sci. 19, 494–506 (2010).

    Article  CAS  Google Scholar 

  53. Sheldrick, G. M. Experimental phasing with SHELXC/D/E: combining chain tracing with density modification. Acta Crystallogr. D Biol. Crystallogr. 66, 479–485 (2010).

    Article  CAS  Google Scholar 

  54. Adams, P. D. et al. PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr. D Biol. Crystallogr. 58, 1948–1954 (2002).

    Article  Google Scholar 

  55. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 (2004).

    Article  Google Scholar 

  56. Abraham, M. J. et al. GROMACS: high-performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1, 19–25 (2015).

    Article  Google Scholar 

  57. PyMOL. The PyMOL Molecular Graphics System, Version 2.4 (Schrodinger Inc., 2015).

  58. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101 (2007).

    Article  Google Scholar 

  59. Wang, H., Dommert, F. & Holm, C. Optimizing working parameters of the smooth particle mesh Ewald algorithm in terms of accuracy and efficiency. J. Chem. Phys. 133, 034117 (2010).

    Article  Google Scholar 

Download references

Acknowledgements

We thank H. Tang in Chu Wang’s lab and the Institute of Geographic Sciences and Natural Resources Research, CAS for the help with ICP measurements. We thank National Center for Protein Sciences at Peking University, Beijing for the help with Circular Dichroism measurements, and the staff of the Shanghai Synchrotron Radiation Facility and KEK Photon Factory for assistance with X-ray data collection. Funding: C.W. was supported by the National Natural Science Foundation of China (grants 21925701, 91953109 and 92153301).

Author information

Author notes

  1. These authors contributed equally: Yao Cheng, Haobo Wang, Hua Xu, Yuan Liu.

Authors and Affiliations

  1. Synthetic and Functional Biomolecules Center, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Bioorganic Chemistry and Molecular Engineering of Ministry of Education, Peking University, Beijing, China

    Yao Cheng, Haobo Wang, Yuan Liu, Bin Ma, Xuemin Chen, Xianghe Wang & Chu Wang

  2. Department of Chemical Biology, College of Chemistry and Molecular Engineering, Peking University, Beijing, China

    Yao Cheng, Haobo Wang, Yuan Liu, Bin Ma, Xuemin Chen, Xianghe Wang & Chu Wang

  3. State Key Laboratory of Protein and Plant Gene Research, and Biomedical Pioneering Innovation Center (BIOPIC), Peking University, Beijing, China

    Hua Xu, Bo Wang & Xiao-Dong Su

  4. Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China

    Xin Zeng & Chu Wang

  5. Cornell University, Ithaca, NY, USA

    Carina Shiau

  6. John Harvard Distinguished Science Fellow, Harvard University, Cambridge, MA, USA

    Sergey Ovchinnikov

Authors
  1. Yao Cheng
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Haobo Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Hua Xu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Yuan Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Bin Ma
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Xuemin Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Xin Zeng
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Xianghe Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Bo Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Carina Shiau
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Sergey Ovchinnikov
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Xiao-Dong Su
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Chu Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

H.W. and C.W. conceived the project; Y.C., H.W. and Y.L. performed computational analysis with the help of C.S. and S.O.; H.X. purified citX and solved the structure under the guidance of X.S.; B.M., X.C., X.W. and X.Z. contributed to biochemical verification of zinc binding in citX and other proteins; and Y.C, H.W., Y.L. and C.W. analyzed the data and wrote the manuscript with inputs from all authors.

Corresponding authors

Correspondence to Yuan Liu, Xiao-Dong Su or Chu Wang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Chemical Biology thanks Kevin Yang, Rosalin Bonetta Valentino and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Distribution of the metal types of the metalloproteins in the training set.

ZN: zinc; CA: calcium; MG: magnesium; MN: manganese; FE: iron; SF4: [FE4-S4] iron-sulfur clusters; NI: nickel; CU: copper; CO: cobalt; FES: [FE2-S2] iron-sulfur clusters.

Source data

Extended Data Fig. 2 Examples of the coevolved CHED network clusters detected in the PDB data.

Metal-binding residues in these sites have the tendency to coevolve with each other and cluster together in a high-order coevolution network. The PDB ID and type of the metal ion bound are listed below the corresponding coevolved network.

Extended Data Fig. 3 Distribution of the number of coevolved connections (‘node degree’) for each residue among all the coevolved CHED pairs in the training set.

The CHED residues involved in metal binding (blue) have more node degrees on average than those non-metal-binding residues in the coevolved networks (red).

Source data

Extended Data Fig. 4 Composition of metal-chelating CHED sidechains within the coordination sphere around each specific metal ion.

The statistic is calculated based on metalloprotein structures in the benchmark. Figure a and b show the absolute count and normalized percentage, respectively. ZN: zinc; CA: calcium; MG: magnesium; MN: manganese; NI: nickel; FE: iron; CU: copper; SF4: [FE4-S4] iron-sulfur clusters; CO: cobalt; FES: [FE2-S2] iron-sulfur clusters.

Source data

Extended Data Fig. 5 Overall workflow of MetalNet.

Starting with the information from coevolution analysis, MetalNet uses a machine learning (ML)-based classifier that has been trained by a benchmark of known metalloproteins to predict whether an individual CHED coevolved residue pair is metal-binding or not. It then employs a graph-based approach to identify high-order coevolution network clusters formed by these coevolved CHED pairs to generate reliable predictions of metal-binding sites in the query protein. In certain cases when the coevolution network topology can be matched to that of known metal-binding sites, the method can also infer the information on the type of metal bound in the predicted site. The method does not use any sequence (1D) or structural (3D) homology information to make predictions.

Extended Data Fig. 6 Comparation of the performance between MetalNet and MIB.

a, Evaluation of the performance of MetalNet and MIB on a ‘prospective’ metalloprotein dataset that were deposited in PDB after 2016/08. b, Evaluation of the performance of MetalNet and MIB on this prospective metalloprotein dataset after proteins with one single sidechain as the liganding group were removed. In a and b, the calculated precision, recall and F1-scores for MetalNet and MIB were shown on the left in the table format for all metal-binding sites (‘All sites’), Zn-specific sites (‘Zn’), Mg/Ca-specific sites (‘Mg/Ca’) and the remaining sites (‘others’). The number of correct and incorrect predictions made by MetalNet and MIB were shown on the right in the venn diagram format with the ‘prospective’ metalloprotein dataset (‘PDB’) as the reference (the dataset can be found in Supplementary Dataset 3).

Source data

Extended Data Fig. 7 Distribution of the minimal distances between the metal-binding residue pairs predicted by MetalNet.

The blue block shows the distribution of residue-residue pairwise distances for metal-binding residues from known metalloproteins in the PDB database. Black solid lines show the distribution of mean distances in GREMLIN models for the metal-binding coevolved sites predicted by MetalNet. The distributions generally agree well with each other until 4.5 Å, suggesting that a large portion of MetalNet predictions indeed have metal-binding residues in proximity with each other. Since GREMLIN does not take metal-binding into consideration during structural modeling, the disagreement after 4.5 Å may come from either false positive prediction by MetalNet, or some errors in structural modeling by GREMLIN.

Source data

Extended Data Fig. 8 Validation of metal binding in purified Desor_0198 (a), SVEN_5263 (b), CbrC (c) and CitX (d) by ICP-MS.

For a and b, the coevolved metal-binding cluster predicted by MetalNet are shown on the left, SDS-PAGE of the purified wild-type protein in the middle and ICP-MS measurement of metal binding of the purified protein are shown on the right. Both proteins are validated with zinc-binding activity. c, The coevolved metal-binding cluster predicted by MetalNet are shown on the left, SDS-PAGE of the purified wild-type protein, two single mutants (C56S or C182S) and the double mutant (C56S&C182S) in the middle and ICP-MS analysis of zinc binding in the purified wild-type cbrC and mutants on the right. Partial zinc binding was retained in each of the single mutant whereas metal binding activity was completely abolished in the double mutant. In a, b, and c, error bars mean and s.d (n = 3 biologically independent samples). ICP-MS analysis of Fe and Cu binding is measured only once. d, SDS-PAGE of the purified wild-type citX as well as four single mutants of the predicted metal-binding residues by MetalNet (C145S, C148S, C155S and H161S). ICP-MS analysis of citX is shown in Fig. 4d. The experiment was repeated twice independently with similar results.

Source data

Extended Data Fig. 9 Molecular dynamic (MD) simulations of citX and its mutants.

MD simulations were performed to calculate root mean square fluctuation (RMSF) for citX (PDB ID: 7DCN) and its mutants (C145S, C148S, C155S and H161, metal-binding residues predicted by MetalNet) using GROMACS. The results suggested that all mutants show greater conformational fluctuations at the binding site.

Source data

Extended Data Fig. 10 Prediction of metal-binding sites in the human spliceosome.

a, Scheme of predicting metal-binding sites in the human spliceosome using coevolution obtained by deepMSA and MSA transformer. b, Highlight of the zinc-binding sites in the human spliceosome predicted by MetalNet that match well with the experimental structures in PDB. The experimental structures of the corresponding metalloproteins subunits are shown in cartoon on the left and the predicted coevolution networks corresponding to the metal-binding sites are shown on the right. Metal-chelating residues are shown in sticks and zinc ions are shown in spheres.

Supplementary information

Supplementary Information

Supplementary Tables 1–5 and Supplementary Fig. 1.

Reporting Summary

Supplementary Data 1

The list of co-evolved pairs used for training and evaluating of ML model in this study.

Supplementary Data 2

The adjacency list of ‘motifs‘ in co-evolution motif bank.

Supplementary Data 3

The ‘prospective‘ metalloprotein dataset that was deposited in PDB after August 2016. It was used for comparing the performance of MetalNet and MIB in an unbiased manner.

Supplementary Data 4

The predicted metal-binding sites by MetalNet in the prokaryotic species dataset.

Supplementary Data 5

The predicted metal-binding sites in metagenome-Pfam dataset.

Supplementary Data 6

The predicted metal-binding sites in the human spliceosome sequences dataset.

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 3

Statistical source data.

Source Data Fig. 4

Unprocessed gels and statistical source data.

Source Data Extended Data Fig. 1

Statistical source data.

Source Data Extended Data Fig. 3

Statistical source data.

Source Data Extended Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 8

Unprocessed gels and ICP-MS data.

Source Data Extended Data Fig. 9

MD data.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cheng, Y., Wang, H., Xu, H. et al. Co-evolution-based prediction of metal-binding sites in proteomes by machine learning. Nat Chem Biol (2023). https://doi.org/10.1038/s41589-022-01223-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s41589-022-01223-z

Search

Advanced search

Quick links

Nature Briefing: Translational Research

Sign up for the Nature Briefing: Translational Research newsletter — top stories in biotechnology, drug discovery and pharma.

Get what matters in translational research, free to your inbox weekly. Sign up for Nature Briefing: Translational Research