D-SCRIPT translates genome to phenome with sequence-based, structure-aware, genome-scale predictions of protein-protein interactions

Sledzieski, Samuel; Singh, Rohit; Cowen, Lenore; Berger, Bonnie

doi:10.1016/j.cels.2021.08.010

Cited by 101 publications

(175 citation statements)

References 104 publications

Supporting

Mentioning

150

Contrasting

Order By: Relevance

“…We showed here the limits of classic sequence-based deep learning models for cross-species predictions, but it is worth noting some recent deep learning models that have been successfully used for cross-species predictions [46], [47] by including biological and chemical information about amino acids as well as structural knowledge. The results presented in this work can hopefully guide similar future work and help move this area further.…”

Section: Discussionmentioning

confidence: 97%

Pitfalls of machine learning models for protein-protein interactions

Lannelongue

Inouye

2022

Preprint

View full text Add to dashboard Cite

Protein-protein interactions (PPIs) are essential to understanding biological pathways as well as their roles in development and disease. Computational tools have been successful at predicting PPIs in silico, but the lack of consistent and reliable frameworks for this task has led to network models that are difficult to compare and, overall, a low level of trust in the PPI predictions. To better understand the underlying mechanisms that underpin these models, we designed B4PPI, an open-source framework for benchmarking that accounts for a range of biological and statistical pitfalls while facilitating reproducibility. We use B4PPI to shed light on the impact of network topology and how different algorithms deal with highly connected proteins. By studying functional genomics-based and sequence-based models (the two most popular approaches) on human PPIs, we show their complementarity as the former performs best on lone proteins while the latter specialises in interactions involving hubs. We also show that algorithm design has little impact on performance with functional genomic data. We replicate our results between both human and S. Cerevisiae data and demonstrate that models using functional genomics are better suited to PPI prediction across species. With rapidly increasing amounts of sequence and functional genomics data, our study provides a systematic foundation for future construction, comparison and application of PPI networks.

show abstract

Section: Discussionmentioning

confidence: 97%

Pitfalls of machine learning models for protein-protein interactions

Lannelongue

Inouye

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In the case of GPCRs, we looked at specificity determining positions (Capra & Singh, 2007;Kalinina et al, 2004), which can then be confirmed by doing 3D structural modeling, to check that the active sites and binding pockets that are expected, should the functional role of the protein be conserved, are indeed present. In the case of G proteins, we derived function based on protein-protein interactions, where we have used our recent deep learning method (Sledzieski et al, 2021) to perform in silico mutagenesis studies to further help us distinguish sequences which are likely to allow us to correctly transfer functional annotation from their human homologs from those that may have other functions. In the case of TLRs, we have used PROSITE to identify presence or absence of entire domains within the sequences, and used expert knowledge to evaluate if the location of these domains matches the functional expectation (e.g.…”

Section: Discussionmentioning

confidence: 99%

“…We took the union of top hits to identify 124 candidate alpha proteins, 207 candidate beta proteins, and 5 candidate gamma proteins. We used the human pre-trained D-SCRIPT (Sledzieski et al, 2021) model to predict interaction between all pairs of alpha-beta, beta-gamma, and alpha-gamma subunits. We performed the same analysis in Montipora capitata (Shumaker et al, 2019), where we identified 184 candidate alpha proteins, 253 candidate beta proteins, and 4 candidate gamma proteins.…”

Section: (D) Transmembrane Helix Detectionmentioning

confidence: 99%

“…(ii) Network analysis of putative P. damicornis G protein subunits: To overcome the challenge that sequence alignment alone was not sufficient to narrow down the choices of P. damicornis sequences related to G proteins, we leveraged computational prediction of protein-protein interactions (PPI) to characterize the relative likelihood of the top P. damicornis hits being truly involved in GPCR binding activity. Applying D-SCRIPT (Sledzieski et al, 2021), a recently introduced sequence-based deep learning model for PPI prediction, we performed two complementary PPI analyses. In the first, we performed an all-vs-all computational screen of interaction between all the candidate G-alpha, beta and gamma proteins in P. damicornis, reasoning that the true positive hits will display the expected interaction patterns of alpha-beta and beta-gamma binding.…”

Section: (C) Gpcr Binding Partner: G Proteinsmentioning

confidence: 99%

See 1 more Smart Citation

Transfer of Knowledge from Model Organisms to Evolutionarily Distant Non-Model Organisms: The CoralPocillopora damicornisMembrane Signaling Receptome

Kumar

Brenner

Sledieski

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

With the ease of gene sequencing and the technology available to study and manipulate non-model organisms, the need to translate our understanding of model organisms to non-model organisms has become an urgent problem. For example, mining of large coral and their symbiont sequence data is a challenge, but also provides an opportunity for understanding functionality and evolution of these and other non-model organisms. Much more information than for any other eukaryotic species is available for humans, especially related to signal transduction and diseases. However, the coral cnidarian host and human have diverged over 700 million years ago and homologies between proteins are therefore often in the gray zone or undetectable with traditional BLAST searches. We introduce a two-stage approach to identifying putative coral homologues of human proteins. First, through remote homology detection using Hidden Markov Models, we identify candidate human homologues in the cnidarian genome. However, for many proteins, the human genome alone contains multiple family members with similar or even more divergence in sequence. In the second stage, therefore, we filter the remote homology results based on the functional and structural plausibility of each coral candidate, shortlisting the coral proteins likely to be true human homologues. We demonstrate our approach with a pipeline for mapping membrane receptors in humans to membrane receptors in corals, with specific focus on the stony coral, P. damicornis. More than 1000 human membrane receptors mapped to 335 coral receptors, including 151 G protein coupled receptors (GPCRs). To validate specific sub-families, we chose opsin proteins, representative GPCRs that confer light sensitivity, and Toll-like receptors, representative non-GPCRs, which function in the immune response, and their ability to communicate with microorganisms. Through detailed structure-function analysis of their ligand-binding pockets and downstream signaling cascades, we selected those candidate remote homologues likely to carry out related functions in the corals. This pipeline may prove generally useful for other non-model organisms, such as to support the growing field of synthetic biology.

show abstract

“…Prediction of PPIs is a non-trivial task. Recent tools (Chen et al ., 2019; Hashemifar et al ., 2018) have been found to lack generalisability in predicting PPIs across species (Sledzieski et al ., 2021). Furthermore, most tools were trained and evaluated using specific studies/databases, raising a question of whether these tools are generalisable across studies/databases.…”

Section: Introductionmentioning

confidence: 99%

LazyPair: scalable prediction of protein-protein interactions and interaction types

Lim

Bhandari

Gardner

2022

Preprint

View full text Add to dashboard Cite

MotivationAlmost all cellular processes require protein-protein interactions. Common interaction types include binding, post-translational modifications, and catalysis. However, existing prediction tools do not take these interaction types into account and do not scale well on proteome-wide prediction.ResultsHere we show that a random forest classifier trained on per-residue physicochemical and biochemical properties is useful for predicting protein-protein interactions. Counterintuitively, we find that training random forests by individual interaction types improves accuracy. Furthermore, a combination of these specialised classifiers improves generalisability. We call our protein-protein interaction prediction tool LazyPair. More importantly, LazyPair outperforms the state-of-the-art in accuracy, generalisability and scalability.Availability and implementationLazyPair and the source code and data for reproducing our analysis are freely available at https://github.com/Gardner-BinfLab/PPI_Analysis_2022 and https://doi.org/10.5281/zenodo.6071630. The web server version and the source code are freely available at https://tisigner.com/lazypair/ and https://github.com/Gardner-BinfLab/TISIGNER-ReactJS, respectively.

show abstract

D-SCRIPT translates genome to phenome with sequence-based, structure-aware, genome-scale predictions of protein-protein interactions

Cited by 101 publications

References 104 publications

Pitfalls of machine learning models for protein-protein interactions

Pitfalls of machine learning models for protein-protein interactions

Transfer of Knowledge from Model Organisms to Evolutionarily Distant Non-Model Organisms: The CoralPocillopora damicornisMembrane Signaling Receptome

LazyPair: scalable prediction of protein-protein interactions and interaction types

Contact Info

Product

Resources

About