HeteroTCR: A heterogeneous graph neural network-based method for predicting peptide-TCR interaction.

Identifying interactions between T-cell receptors (TCRs) and immunogenic peptides holds profound implications across diverse research domains and clinical scenarios. Unsupervised clustering models (UCMs) cannot predict peptide-TCR binding directly, while supervised predictive models (SPMs) often face challenges in identifying antigens previously unencountered by the immune system or possessing limited TCR binding repertoires. Therefore, we propose HeteroTCR, an SPM based on Heterogeneous Graph Neural Network (GNN), to accurately predict peptide-TCR binding probabilities. HeteroTCR captures within-type (TCR-TCR or peptide-peptide) similarity information and between-type (peptide-TCR) interaction insights for predictions on unseen peptides and TCRs, surpassing limitations of existing SPMs. Our evaluation shows HeteroTCR outperforms state-of-the-art models on independent datasets. Ablation studies and visual interpretation underscore the Heterogeneous GNN module's critical role in enhancing HeteroTCR's performance by capturing pivotal binding process features. We further demonstrate the robustness and reliability of HeteroTCR through validation using single-cell datasets, aligning with the expectation that pMHC-TCR complexes with higher predicted binding probabilities correspond to increased binding fractions.

TULIP: A transformer-based unsupervised language model for interacting peptides and T cell receptors that generalizes to unseen epitopes.

The accurate prediction of binding between T cell receptors (TCR) and their cognate epitopes is key to understanding the adaptive immune response and developing immunotherapies. Current methods face two significant limitations: the shortage of comprehensive high-quality data and the bias introduced by the selection of the negative training data commonly used in the supervised learning approaches. We propose a method, Transformer-based Unsupervised Language model for Interacting Peptides and T cell receptors (TULIP), that addresses both limitations by leveraging incomplete data and unsupervised learning and using the transformer architecture of language models. Our model is flexible and integrates all possible data sources, regardless of their quality or completeness. We demonstrate the existence of a bias introduced by the sampling procedure used in previous supervised approaches, emphasizing the need for an unsupervised approach. TULIP recognizes the specific TCRs binding an epitope, performing well on unseen epitopes. Our model outperforms state-of-the-art models and offers a promising direction for the development of more accurate TCR epitope recognition models.

Calcium polypeptide mitigates Cd toxicity in rice via reducing oxidative stress and regulating pectin modification.

KEY MESSAGE: Calcium polypeptide plays a key role during cadmium stress responses in rice, which is involved in increasing peroxidase activity, modulating pectin methylesterase activity, and regulating cell wall by reducing malondialdehyde content. Cadmium (Cd) contamination threatens agriculture and human health globally, emphasizing the need for sustainable methods to reduce cadmium toxicity in crops. Calcium polypeptide (CaP) is a highly water-soluble small molecular peptide acknowledged for its potential as an organic fertilizer in promoting plant growth. However, it is still unknown whether CaP has effects on mitigating Cd toxicity. Here, we investigated the effect of CaP application on the ability to tolerate toxic Cd in rice. We evaluated the impact of CaP on rice seedlings under varying Cd stress conditions and investigated the effect mechanism of CaP mitigating Cd toxicity by Fourier transform infrared spectroscopy (FTIR), fluorescent probe dye, immunofluorescent labeling, and biochemical analysis. We found a notable alleviation of Cd toxicity by reduced malondialdehyde content and increased peroxidase activity. In addition, our findings reveal that CaP induces structural alterations in the root cell wall by modulating pectin methylesterase activity. Altogether, our results confirm that CaP not only promoted biomass accumulation but also reduced Cd concentration in rice. This study contributes valuable insights to sustainable strategies for addressing Cd contamination in agricultural ecosystems.

Highly biased agonism for GPCR ligands via nanobody tethering.

Ligand-induced activation of G protein-coupled receptors (GPCRs) can initiate signaling through multiple distinct pathways with differing biological and physiological outcomes. There is intense interest in understanding how variation in GPCR ligand structure can be used to promote pathway selective signaling ("biased agonism") with the goal of promoting desirable responses and avoiding deleterious side effects. Here we present an approach in which a conventional peptide ligand for the type 1 parathyroid hormone receptor (PTHR1) is converted from an agonist which induces signaling through all relevant pathways to a compound that is highly selective for a single pathway. This is achieved not through variation in the core structure of the agonist, but rather by linking it to a nanobody tethering agent that binds with high affinity to a separate site on the receptor not involved in signal transduction. The resulting conjugate represents the most biased agonist of PTHR1 reported to date. This approach holds promise for facile generation of pathway selective ligands for other GPCRs.

Inhibition and transport mechanisms of the ABC transporter hMRP5.

Human multidrug resistance protein 5 (hMRP5) effluxes anticancer and antivirus drugs, driving multidrug resistance. To uncover the mechanism of hMRP5, we determine six distinct cryo-EM structures, revealing an autoinhibitory N-terminal peptide that must dissociate to permit subsequent substrate recruitment. Guided by these molecular insights, we design an inhibitory peptide that could block substrate entry into the transport pathway. We also identify a regulatory motif, comprising a positively charged cluster and hydrophobic patches, within the first nucleotide-binding domain that modulates hMRP5 localization by engaging with membranes. By integrating our structural, biochemical, computational, and cell biological findings, we propose a model for hMRP5 conformational cycling and localization. Overall, this work provides mechanistic understanding of hMRP5 function, while informing future selective hMRP5 inhibitor development. More broadly, this study advances our understanding of the structural dynamics and inhibition of ABC transporters.

A production platform for disulfide-bonded peptides in the periplasm of Escherichia coli.

BACKGROUND: Recombinant peptide production in Escherichia coli provides a sustainable alternative to environmentally harmful and size-limited chemical synthesis. However, in-vivo production of disulfide-bonded peptides at high yields remains challenging, due to degradation by host proteases/peptidases and the necessity of translocation into the periplasmic space for disulfide bond formation. RESULTS: In this study, we established an expression system for efficient and soluble production of disulfide-bonded peptides in the periplasm of E. coli. We chose model peptides with varying complexity (size, structure, number of disulfide bonds), namely parathyroid hormone 1-84, somatostatin 1-28, plectasin, and bovine pancreatic trypsin inhibitor (aprotinin). All peptides were expressed without and with the N-terminal, low molecular weight CASPON™ tag (4.1 kDa), with the expression cassette being integrated into the host genome. During BioLector™ cultivations at microliter scale, we found that most of our model peptides can only be sufficiently expressed in combination with the CASPON™ tag, otherwise expression was only weak or undetectable on SDS-PAGE. Undesired degradation by host proteases/peptidases was evident even with the CASPON™ tag. Therefore, we investigated whether degradation happened before or after translocation by expressing the peptides in combination with either a co- or post-translational signal sequence. Our results suggest that degradation predominantly happened after the translocation, as degradation fragments appeared to be identical independent of the signal sequence, and expression was not enhanced with the co-translational signal sequence. Lastly, we expressed all CASPON™-tagged peptides in two industry-relevant host strains during C-limited fed-batch cultivations in bioreactors. We found that the process performance was highly dependent on the peptide-host-combination. The titers that were reached varied between 0.6-2.6 g L-1, and exceeded previously published data in E. coli. Moreover, all peptides were shown by mass spectrometry to be expressed to completion, including full formation of disulfide bonds. CONCLUSION: In this work, we demonstrated the potential of the CASPON™ technology as a highly efficient platform for the production of soluble peptides in the periplasm of E. coli. The titers we show here are unprecedented whenever parathyroid hormone, somatostatin, plectasin or bovine pancreatic trypsin inhibitor were produced in E. coli, thus making our proposed upstream platform favorable over previously published approaches and chemical synthesis.

A novel ACE inhibitory peptide from Pelodiscus sinensis Wiegmann meat water-soluble protein hydrolysate.

Pelodiscus sinensis meat is a nutritional food and tonic with angiotensin-converting enzyme (ACE) inhibitory activities. To identify the bioactive substances responsible, several bioinformatics methods were integrated to enable a virtual screening for bioactive peptides in proteins identified within a water-soluble protein fraction of Pelodiscus sinensis meat by Shotgun proteomics. The peptides were generated from the identified proteins by in silico proteolysis using six proteases. A comparison of the numbers of proteins suitable for digestion with each enzyme and the iBAQ (intensity-based absolute quantification) values for these proteins revealed that bromelain and papain were the most suitable proteases for this sample. Next, the water solubility, toxicity, and ADMET (absorption/distribution/metabolism/excretion/toxicity) properties of these peptides were evaluated in silico. Finally, a novel ACE inhibitory peptide IEWEF with an IC50 value of 41.33 µM was identified. The activity of the synthesized peptide was verified in vitro, and it was shown to be a non-competitive ACE inhibitor. Molecular docking revealed that IEWEF could tightly bind to C-ACE, and N-ACE with energies less than 0 kJ mol-1, and the peptide IEWEF can form hydrogen bonds with C-ACE and N-ACE respectively. These results provide evidence that bioactive peptides in the water-soluble protein fraction account for (at least) some of the ACE inhibitory activities observed in Pelodiscus sinensis meat. Furthermore, our research provides a workflow for the efficient identification of novel ACE inhibitory peptides from complex protein mixtures.

Principles of peptide selection by the transporter associated with antigen processing.

Our ability to fight pathogens relies on major histocompatibility complex class I (MHC-I) molecules presenting diverse antigens on the surface of diseased cells. The transporter associated with antigen processing (TAP) transports nearly the entire repertoire of antigenic peptides into the endoplasmic reticulum for MHC-I loading. How TAP transports peptides specific for MHC-I is unclear. In this study, we used cryo-EM to determine a series of structures of human TAP, both in the absence and presence of peptides with various sequences and lengths. The structures revealed that peptides of eight or nine residues in length bind in a similarly extended conformation, despite having little sequence overlap. We also identified two peptide-anchoring pockets on either side of the transmembrane cavity, each engaging one end of a peptide with primarily main chain atoms. Occupation of both pockets results in a global conformational change in TAP, bringing the two halves of the transporter closer together to prime it for isomerization and ATP hydrolysis. Shorter peptides are able to bind to each pocket separately but are not long enough to bridge the cavity to bind to both simultaneously. Mutations that disrupt hydrogen bonds with the N and C termini of peptides almost abolish MHC-I surface expression. Our findings reveal that TAP functions as a molecular caliper that selects peptides according to length rather than sequence, providing antigen diversity for MHC-I presentation.

Drosophila in the study of hTBP protein interactions in the development and modeling of SCA17.

BACKGROUND: Protein interactions participate in many molecular mechanisms involved in cellular processes. The human TATA box binding protein (hTBP) interacts with Antennapedia (Antp) through its N-terminal region, specifically via its glutamine homopeptides. This PolyQ region acts as a binding site for other transcription factors under normal conditions, but when it expands, it generates spinocerebellar ataxia 17 (SCA17), whose protein aggregates in the brain prevent its correct functioning. OBJECTIVE: To determine whether the hTBP glutamine-rich region is involved in its interaction with homeoproteins and the role it plays in the formation of protein aggregates in SCA17. MATERIAL AND METHODS: We characterized hTBP interaction with other homeoproteins using BiFC, and modeled SCA17 in Drosophila melanogaster by targeting hTBPQ80 to the fly brain using UAS/GAL4. RESULTS: There was hTBP interaction with homeoproteins through its glutamine-rich region, and hTBP protein aggregates with expanded glutamines were found to affect the locomotor capacity of flies. CONCLUSIONS: The study of hTBP interactions opens the possibility for the search for new therapeutic strategies in neurodegenerative pathologies such as SCA17.

ANTECEDENTES: Las interacciones proteicas participan en una gran cantidad de mecanismos moleculares que rigen los procesos celulares. La proteína de unión a la caja TATA humana (hTBP) interacciona con Antennapedia (Antp) a través de su extremo N-terminal, específicamente a través de sus homopéptidos de glutaminas. Esta región PolyQ sirve como sitio de unión a factores de transcripción en condiciones normales, pero cuando se expande genera la ataxia espinal cerebelosa 17 (SCA17), cuyos agregados proteicos en el cerebro impiden su funcionamiento correcto. OBJETIVO: Determinar si la región rica en glutaminas de hTBP interviene en su interacción con homeoproteínas y el papel que tiene en la formación de agregados proteicos en SCA17. MATERIAL Y MÉTODOS: Se caracterizó la interacción de hTBP con otras homeoproteínas usando BiFC y se modeló SCA17 en Drosophila melanogaster dirigiendo hTBPQ80 al cerebro de las moscas usando UAS/GAL4. RESULTADOS: Existió interacción de hTBP con homeoproteínas a través de su región rica en glutaminas. Los agregados proteicos de hTBP con las glutaminas expandidas afectaron la capacidad locomotriz de las moscas. CONCLUSIONES: El estudio de las interacciones de hTBP abre la posibilidad para la búsqueda de nuevas estrategias terapéuticas en patologías neurodegenerativas como SCA17.

Autophagy preferentially degrades non-fibrillar polyQ aggregates.

Aggregation of proteins containing expanded polyglutamine (polyQ) repeats is the cytopathologic hallmark of a group of dominantly inherited neurodegenerative diseases, including Huntington's disease (HD). Huntingtin (Htt), the disease protein of HD, forms amyloid-like fibrils by liquid-to-solid phase transition. Macroautophagy has been proposed to clear polyQ aggregates, but the efficiency of aggrephagy is limited. Here, we used cryo-electron tomography to visualize the interactions of autophagosomes with polyQ aggregates in cultured cells in situ. We found that an amorphous aggregate phase exists next to the radially organized polyQ fibrils. Autophagosomes preferentially engulfed this amorphous material, mediated by interactions between the autophagy receptor p62/SQSTM1 and the non-fibrillar aggregate surface. In contrast, amyloid fibrils excluded p62 and evaded clearance, resulting in trapping of autophagic structures. These results suggest that the limited efficiency of autophagy in clearing polyQ aggregates is due to the inability of autophagosomes to interact productively with the non-deformable, fibrillar disease aggregates.

Substrate displacement of CK1 C-termini regulates kinase specificity.

CK1 kinases participate in many signaling pathways, and their regulation is of meaningful biological consequence. CK1s autophosphorylate their C-terminal noncatalytic tails, and eliminating these tails increases substrate phosphorylation in vitro, suggesting that the autophosphorylated C-termini act as inhibitory pseudosubstrates. To test this prediction, we comprehensively identified the autophosphorylation sites on Schizosaccharomyces pombe Hhp1 and human CK1ε. Phosphoablating mutations increased Hhp1 and CK1ε activity toward substrates. Peptides corresponding to the C-termini interacted with the kinase domains only when phosphorylated, and substrates competitively inhibited binding of the autophosphorylated tails to the substrate binding grooves. Tail autophosphorylation influenced the catalytic efficiency with which CK1s targeted different substrates, and truncating the tail of CK1δ broadened its linear peptide substrate motif, indicating that tails contribute to substrate specificity as well. Considering autophosphorylation of both T220 in the catalytic domain and C-terminal sites, we propose a displacement specificity model to describe how autophosphorylation modulates substrate specificity for the CK1 family.

Bioinformatics leading to conveniently accessible, helix enforcing, bicyclic ASX motif mimics (BAMMs).

Helix mimicry provides probes to perturb protein-protein interactions (PPIs). Helical conformations can be stabilized by joining side chains of non-terminal residues (stapling) or via capping fragments. Nature exclusively uses capping, but synthetic helical mimics are heavily biased towards stapling. This study comprises: (i) creation of a searchable database of unique helical N-caps (ASX motifs, a protein structural motif with two intramolecular hydrogen-bonds between aspartic acid/asparagine and following residues); (ii) testing trends observed in this database using linear peptides comprising only canonical L-amino acids; and, (iii) novel synthetic N-caps for helical interface mimicry. Here we show many natural ASX motifs comprise hydrophobic triangles, validate their effect in linear peptides, and further develop a biomimetic of them, Bicyclic ASX Motif Mimics (BAMMs). BAMMs are powerful helix inducing motifs. They are synthetically accessible, and potentially useful to a broad section of the community studying disruption of PPIs using secondary structure mimics.

Active droplets through enzyme-free, dynamic phosphorylation.

Life continuously transduces energy to perform critical functions using energy stored in reactive molecules like ATP or NADH. ATP dynamically phosphorylates active sites on proteins and thereby regulates their function. Inspired by such machinery, regulating supramolecular functions using energy stored in reactive molecules has gained traction. Enzyme-free, synthetic systems that use dynamic phosphorylation to regulate supramolecular processes have not yet been reported, to our knowledge. Here, we show an enzyme-free reaction cycle that consumes the phosphorylating agent monoamidophosphate by transiently phosphorylating histidine and histidine-containing peptides. The phosphorylated species are labile and deactivate through hydrolysis. The cycle exhibits versatility and tunability, allowing for the dynamic phosphorylation of multiple precursors with a tunable half-life. Notably, we show the resulting phosphorylated products can regulate the peptide's phase separation, leading to active droplets that require the continuous conversion of fuel to sustain. The reaction cycle will be valuable as a model for biological phosphorylation but can also offer insights into protocell formation.

Membrane-bound chemoreception of bitter bile acids and peptides is mediated by the same subset of bitter taste receptors.

The vertebrate sense of taste allows rapid assessment of the nutritional quality and potential presence of harmful substances prior to ingestion. Among the five basic taste qualities, salty, sour, sweet, umami, and bitter, bitterness is associated with the presence of putative toxic substances and elicits rejection behaviors in a wide range of animals including humans. However, not all bitter substances are harmful, some are thought to be health-beneficial and nutritious. Among those compound classes that elicit a bitter taste although being non-toxic and partly even essential for humans are bitter peptides and L-amino acids. Using functional heterologous expression assays, we observed that the 5 dominant human bitter taste receptors responsive to bitter peptides and amino acids are activated by bile acids, which are notorious for their extreme bitterness. We further demonstrate that the cross-reactivity of bitter taste receptors for these two different compound classes is evolutionary conserved and can be traced back to the amphibian lineage. Moreover, we show that the cross-detection by some receptors relies on "structural mimicry" between the very bitter peptide L-Trp-Trp-Trp and bile acids, whereas other receptors exhibit a phylogenetic conservation of this trait. As some bile acid-sensitive bitter taste receptor genes fulfill dual-roles in gustatory and non-gustatory systems, we suggest that the phylogenetic conservation of the rather surprising cross-detection of the two substance classes could rely on a gene-sharing-like mechanism in which the non-gustatory function accounts for the bitter taste response to amino acids and peptides.

RACER-m leverages structural features for sparse T cell specificity prediction.

Reliable prediction of T cell specificity against antigenic signatures is a formidable task, complicated by the immense diversity of T cell receptor and antigen sequence space and the resulting limited availability of training sets for inferential models. Recent modeling efforts have demonstrated the advantage of incorporating structural information to overcome the need for extensive training sequence data, yet disentangling the heterogeneous TCR-antigen interface to accurately predict MHC-allele-restricted TCR-peptide interactions has remained challenging. Here, we present RACER-m, a coarse-grained structural model leveraging key biophysical information from the diversity of publicly available TCR-antigen crystal structures. Explicit inclusion of structural content substantially reduces the required number of training examples and maintains reliable predictions of TCR-recognition specificity and sensitivity across diverse biological contexts. Our model capably identifies biophysically meaningful point-mutant peptides that affect binding affinity, distinguishing its ability in predicting TCR specificity of point-mutants from alternative sequence-based methods. Its application is broadly applicable to studies involving both closely related and structurally diverse TCR-peptide pairs.

Amino acid insertion in Bat MHC-I enhances complex stability and augments peptide presentation.

Bats serve as reservoirs for numerous zoonotic viruses, yet they typically remain asymptomatic owing to their unique immune system. Of particular significance is the MHC-I in bats, which plays crucial role in anti-viral response and exhibits polymorphic amino acid (AA) insertions. This study demonstrated that both 5AA and 3AA insertions enhance the thermal stability of the bat MHC-I complex and enrich the diversity of bound peptides in terms of quantity and length distribution, by stabilizing the 310 helix, a region prone to conformational changes during peptide loading. However, the mismatched insertion could diminish the stability of bat pMHC-I. We proposed that a suitable insertion may help bat MHC-I adapt to high body temperatures during flight while enhancing antiviral responses. Moreover, this site-specific insertions may represent a strategy of evolutionary adaptation of MHC-I molecules to fluctuations in body temperature, as similar insertions have been found in other lower vertebrates.

Insights on the G protein-coupled receptor helix 8 solution structure and orientation using a neurotensin receptor 1 peptide.

G-protein coupled receptors (GPCRs) are the largest class of membrane proteins encoded in the human genome with high pharmaceutical relevance and implications to human health. These receptors share a prevalent architecture of seven transmembrane helices followed by an intracellular, amphipathic helix 8 (H8) and a disordered C-terminal tail (Ctail). Technological advancements have led to over 1000 receptor structures in the last two decades, yet frequently H8 and the Ctail are conformationally heterogeneous or altogether absent. Here we synthesize a peptide comprising the neurotensin receptor 1 (NTS1) H8 and Ctail (H8-Ctail) to investigate its structural stability, conformational dynamics, and orientation in the presence of detergent and phospholipid micelles, which mimic the membrane. Circular dichroism (CD) and nuclear magnetic resonance (NMR) measurements confirm that zwitterionic 1,2-diheptanoyl-sn-glycero-3-phosphocholine is a potent stabilizer of H8 structure, whereas the commonly-used branched detergent lauryl maltose neopentyl glycol (LMNG) is unable to completely stabilize the helix - even at amounts four orders of magnitude greater than its critical micellar concentration. We then used NMR spectroscopy to assign the backbone chemical shifts. A series of temperature and lipid titrations were used to define the H8 boundaries as F376-R392 from chemical shift perturbations, changes in resonance intensity, and chemical-shift-derived phi/psi angles. Finally, the H8 azimuthal and tilt angles, defining the helix orientation relative of the membrane normal were measured using paramagnetic relaxation enhancement NMR. Taken together, our studies reveal the H8-Ctail region is sensitive to membrane physicochemical properties and is capable of more adaptive behavior than previously suggested by static structural techniques.

Changes in the concentration of bone turnover markers in men after maximum intensity exercise.

Background: Physical activity is an important factor in modelling the remodelling and metabolism of bone tissue. The aim of the study was to evaluate the changes in indices demonstrating bone turnover in men under the influence of maximum-intensity exercise. Methods: The study involved 33 men aged 20-25, divided into two groups: experimental (n = 15) and control (n = 18). People training medium- and long-distance running were assigned to the experimental group, and non-training individuals to the control. Selected somatic, physiological and biochemical indices were measured. The level of aerobic fitness was determined using a progressively increasing graded test (treadmill test for subjective fatigue). Blood samples for determinations were taken before the test and 60 minutes after its completion. The concentration of selected bone turnover markers was assessed: bone fraction of alkaline phosphatase (b-ALP), osteoclacin (OC), N-terminal cross-linked telopeptide of the alpha chain of type I collagen (NTx1), N-terminal propeptide of type I progolagen (PINP), osteoprotegerin (OPG). In addition, the concentration of 25(OH)D3 prior to the stress test was determined. Additionally, pre and post exercise, the concentration of lactates in the capillary blood was determined. Results: When comparing the two groups, significant statistical differences were found for the mean level of: 25(OH)D3 (p = 0.025), b-ALP (p < 0.001), OC (p = 0.004) and PINP (p = 0.029) prior to the test. On the other hand, within individual groups, between the values pre and post the stress test, there were statistically significant differences for the average level of: b-ALP (p < 0.001), NTx1 (p < 0.001), OPG (p = 0.001) and PINP (p = 0.002). Conclusion: A single-session maximum physical effort can become an effective tool to initiate positive changes in bone turnover markers.

Enrichment and Quantitation of Dipeptidyl Peptidase IV Inhibitory Peptides in Quinoa upon Systematic Malting.

Food-derived peptides with an inhibitory effect on dipeptidyl peptidase IV (DPP-IV) can be used as an additive treatment for type 2 diabetes. The inhibitory potential of food depends on technological protein hydrolysis and gastrointestinal digestion, as the peptides only act after intestinal resorption. The effect of malting as a hydrolytic step on the availability of these peptides in grains has yet to be investigated. In this study, quinoa was malted under systematic temperature, moisture, and time variations. In the resulting malts, the DPP-IV inhibition reached a maximum of 45.02 (±10.28) %, whereas the highest overall concentration of literature-known inhibitory peptides was 4.07 µmol/L, depending on the malting parameters. After in vitro gastrointestinal digest, the inhibition of most malts, as well as the overall concentration of inhibitory peptides, could be increased significantly. Additionally, the digested malts showed higher values in both the inhibition and the peptide concentration than the unmalted quinoa. Concerning the malting parameters, germination time had the highest impact on the inhibition and the peptide concentration after digest. An analysis of the protein sizes before and after malting gave first hints toward the origin of these peptides, or their precursors, in quinoa.

Peptriever: a Bi-Encoder approach for large-scale protein-peptide binding search.

MOTIVATION: Peptide therapeutics hinge on the precise interaction between a tailored peptide and its designated receptor while mitigating interactions with alternate receptors is equally indispensable. Existing methods primarily estimate the binding score between protein and peptide pairs. However, for a specific peptide without a corresponding protein, it is challenging to identify the proteins it could bind due to the sheer number of potential candidates. RESULTS: We propose a transformers-based protein embedding scheme in this study that can quickly identify and rank millions of interacting proteins. Furthermore, the proposed approach outperforms existing sequence- and structure-based methods, with a mean AUC-ROC and AUC-PR of 0.73. AVAILABILITY AND IMPLEMENTATION: Training data, scripts, and fine-tuned parameters are available at https://github.com/RoniGurvich/Peptriever. The proposed method is linked with a web application available for customized prediction at https://peptriever.app/.