Evolutionary dynamics of a lattice dimer: a toy model for stability vs. affinity trade-offs in proteins

Loffredo, E; Vesconi, E; Razban, R; Peleg, O; Shakhnovich, E; Cocco, S; Monasson, R

doi:10.1088/1751-8121/acfddc

Cited by 2 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following (47), we build dimeric LPs starting from single monomers running Monte Carlo (MC) evolution, and collect sequence data for multiple dimeric LPs. Spanning multiple dimeric structures, by changing the conformation of the selfavoiding paths on the lattice, allowed us to model different group specificities; details in Methods.…”

Section: Minimal Model Of Out-of-sample Classification Of Tcr-peptidementioning

confidence: 99%

“…Sequence data for LP dimers are obtained from (47), where two monomer sequences v 1 , v 2 -folded in, respectively, structures S 1 , S 2 -form a dimeric complex via the interaction energy [4] where the sum runs over all sites of both structures. The index π in Equation ( 4) labels a specific orientation of the interaction, see (47). In analogy with Equation ( 2), the probability that the sequences v 1 , v 2 fold into the dimer S 1 + S 2 is…”

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

“…[6] The MSAs are collected at steps t = 0, t = 100, t = 500 and represent non binder, weak and strong binder dimers used in this work. We refer the interested reader to (47) for an extensive presentation of the LP model and details on how to obtain sequence data. In particular to generate the synthetic dataset used for the pan-specific model we select 12 pairs of structures and for each of them collect an MSA of binding and non binding sequences with the same length to ensure balance.…”

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

“…The out-of-sample data for close structures is collected by first finding the closest structures to those used in the in-sample dataset by minimizing the mean energy in Equation ( 3) over the in-sample binder MSA across all the structures available in (47,58). We then generate the MSA for the selected close structure pair (2237 + 304), and this constitutes the close out-of-sample dataset.…”

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

“…Benchmarking on synthetic data reveals strong dependence of performance upon out-of-sample data distribution. We take data from (47), where the authors built dimeric LPs starting from single monomers running Monte Carlo (MC) evolution at variable intra-dimer interaction strengths (see Methods). We collect sequence data in the form of Multiple Sequence Alignments (MSA) at three steps during MC evolution, namely at beginning, intermediate and endpoint of evolution: the MSAs will constitute non binder, weak binder and strong binder data, respectively.…”

Section: S2 Out-of-sample Tcr Specificity Predictionsmentioning

confidence: 99%

See 4 more Smart Citations

Restoring data balance via generative models of T-cell receptors for antigen-binding prediction

Loffredo,

Pastore,

Cocco

et al. 2024

Preprint

View full text Add to dashboard Cite

Unveiling the specificity in T-cell-receptor and antigen recognition represents a major step to understand the immune system response. Many supervised machine learning approaches have been designed to build sequence-based predictive models of such specificity using binding and non-binding examples of data. Due to the presence of few specific and many non-specific T-cell receptors for each antigen, available datasets are heavily imbalanced and make the goal of achieving solid predictive performances very challenging. Here, we propose to restore data balance through data augmentation using generative unsupervised models. We then use these augmented data to train supervised models for prediction of peptide-specific T-cell receptors and binding pairs of peptide and T-cell receptors sequences. We show that our pipeline yields increased performance in terms of T-cell receptors specificity prediction tasks. More broadly, our work provides a general framework to restore balance in computational problems involving biological sequence data.Significance StatementThe adaptive immune system carries a diverse set of T-cell receptors capable of recognizing pathogens and protect the host from diseases. Predicting whether a receptor binds a pathogenic peptide is a fundamental computational problem, made difficult by the imbalance in available data: relatively few binding pairs are known compared to all possible pairs of receptors and peptides. Here, we propose to mitigate this imbalance problem by generating putative binding pairs through data augmentation machine-learning methods. We show that these extra data helps training binding prediction models and improves their performances. Our framework for sequence data augmentation is generic and could be applied to other biological computational problems.

show abstract

Section: Minimal Model Of Out-of-sample Classification Of Tcr-peptidementioning

confidence: 99%

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

Section: Synthetic Lattice Protein Dimer Data In This Work We Considermentioning

confidence: 99%

Section: S2 Out-of-sample Tcr Specificity Predictionsmentioning

confidence: 99%

See 3 more Smart Citations

Restoring data balance via generative models of T-cell receptors for antigen-binding prediction

Loffredo,

Pastore,

Cocco

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

Functional effects of mutations in proteins can be predicted and interpreted by guided selection of sequence covariation information

Cocco,

Posani,

Monasson

2024

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

Predicting the effects of one or more mutations to the in vivo or in vitro properties of a wild-type protein is a major computational challenge, due to the presence of epistasis, that is, of interactions between amino acids in the sequence. We introduce a computationally efficient procedure to build minimal epistatic models to predict mutational effects by combining evolutionary (homologous sequence) and few mutational-scan data. Mutagenesis measurements guide the selection of links in a sparse graphical model, while the parameters on the nodes and the edges are inferred from sequence data. We show, on 10 mutational scans, that our pipeline exhibits performances comparable to state-of-the-art deep networks trained on many more data, while requiring much less parameters and being hence more interpretable. In particular, the identified interactions adapt to the wild-type protein and to the fitness or biochemical property experimentally measured, mostly focus on key functional sites, and are not necessarily related to structural contacts. Therefore, our method is able to extract information relevant for one mutational experiment from homologous sequence data reflecting the multitude of structural and functional constraints acting on proteins throughout evolution.

show abstract

Evolutionary dynamics of a lattice dimer: a toy model for stability vs. affinity trade-offs in proteins

Cited by 2 publications

References 33 publications

Restoring data balance via generative models of T-cell receptors for antigen-binding prediction

Restoring data balance via generative models of T-cell receptors for antigen-binding prediction

Functional effects of mutations in proteins can be predicted and interpreted by guided selection of sequence covariation information

Contact Info

Product

Resources

About