Improved deep learning prediction of antigen-antibody interactions.

Identifying antibodies that neutralize specific antigens is crucial for developing effective immunotherapies, but this task remains challenging for many target antigens. The rise of deep learning-based computational approaches presents a promising avenue to address this challenge. Here, we assess the performance of a deep learning approach through two benchmark tests aimed at predicting antibodies for the receptor-binding domain of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike protein. Three different strategies for constructing input sequence alignments are employed for predicting structural models of antigen-antibody complexes. In our initial testing set, which comprises known experimental structures, these strategies collectively yield a significant top-ranked prediction for 61% of cases and a success rate of 47%. Notably, one strategy that utilizes the sequences of known antigen binders outperforms the other two, achieving a precision of 90% in a subsequent test set of ~1,000 antibodies, balanced between true and control antibodies for the antigen, albeit with a lower recall of 25%. Our results underscore the potential of integrating deep learning methods with single B cell sequencing techniques to enhance the prediction accuracy of antigen-antibody interactions.

Predicted structural proteome of Sphagnum divinum and proteome-scale annotation.

MOTIVATION: Sphagnum-dominated peatlands store a substantial amount of terrestrial carbon. The genus is undersampled and under-studied. No experimental crystal structure from any Sphagnum species exists in the Protein Data Bank and fewer than 200 Sphagnum-related genes have structural models available in the AlphaFold Protein Structure Database. Tools and resources are needed to help bridge these gaps, and to enable the analysis of other structural proteomes now made possible by accurate structure prediction. RESULTS: We present the predicted structural proteome (25 134 primary transcripts) of Sphagnum divinum computed using AlphaFold, structural alignment results of all high-confidence models against an annotated nonredundant crystallographic database of over 90,000 structures, a structure-based classification of putative Enzyme Commission (EC) numbers across this proteome, and the computational method to perform this proteome-scale structure-based annotation. AVAILABILITY AND IMPLEMENTATION: All data and code are available in public repositories, detailed at https://github.com/BSDExabio/SAFA. The structural models of the S. divinum proteome have been deposited in the ModelArchive repository at https://modelarchive.org/doi/10.5452/ma-ornl-sphdiv.

A personalized probabilistic approach to ovarian cancer diagnostics.

OBJECTIVE: The identification/development of a machine learning-based classifier that utilizes metabolic profiles of serum samples to accurately identify individuals with ovarian cancer. METHODS: Serum samples collected from 431 ovarian cancer patients and 133 normal women at four geographic locations were analyzed by mass spectrometry. Reliable metabolites were identified using recursive feature elimination coupled with repeated cross-validation and used to develop a consensus classifier able to distinguish cancer from non-cancer. The probabilities assigned to individuals by the model were used to create a clinical tool that assigns a likelihood that an individual patient sample is cancer or normal. RESULTS: Our consensus classification model is able to distinguish cancer from control samples with 93% accuracy. The frequency distribution of individual patient scores was used to develop a clinical tool that assigns a likelihood that an individual patient does or does not have cancer. CONCLUSIONS: An integrative approach using metabolomic profiles and machine learning-based classifiers has been employed to develop a clinical tool that assigns a probability that an individual patient does or does not have ovarian cancer. This personalized/probabilistic approach to cancer diagnostics is more clinically informative and accurate than traditional binary (yes/no) tests and represents a promising new direction in the early detection of ovarian cancer.

A novel sequence alignment algorithm based on deep learning of the protein folding code.

MOTIVATION: From evolutionary interference, function annotation to structural prediction, protein sequence comparison has provided crucial biological insights. While many sequence alignment algorithms have been developed, existing approaches often cannot detect hidden structural relationships in the 'twilight zone' of low sequence identity. To address this critical problem, we introduce a computational algorithm that performs protein Sequence Alignments from deep-Learning of Structural Alignments (SAdLSA, silent 'd'). The key idea is to implicitly learn the protein folding code from many thousands of structural alignments using experimentally determined protein structures. RESULTS: To demonstrate that the folding code was learned, we first show that SAdLSA trained on pure α-helical proteins successfully recognizes pairs of structurally related pure ß-sheet protein domains. Subsequent training and benchmarking on larger, highly challenging datasets show significant improvement over established approaches. For challenging cases, SAdLSA is ∼150% better than HHsearch for generating pairwise alignments and ∼50% better for identifying the proteins with the best alignments in a sequence library. The time complexity of SAdLSA is O(N) thanks to GPU acceleration. AVAILABILITY AND IMPLEMENTATION: Datasets and source codes of SAdLSA are available free of charge for academic users at http://sites.gatech.edu/cssb/sadlsa/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

On the possible origin of protein homochirality, structure, and biochemical function.

Living systems have chiral molecules, e.g., native proteins that almost entirely contain L-amino acids. How protein homochirality emerged from a background of equal numbers of L and D amino acids is among many questions about life's origin. The origin of homochirality and its implications are explored in computer simulations examining the stability and structural and functional properties of an artificial library of compact proteins containing 1:1 (termed demi-chiral), 3:1, and 1:3 ratios of D:L and purely L or D amino acids generated without functional selection. Demi-chiral proteins have shorter secondary structures and fewer internal hydrogen bonds and are less stable than homochiral proteins. Selection for hydrogen bonding yields a preponderance of L or D amino acids. Demi-chiral proteins have native global folds, including similarity to early ribosomal proteins, similar small molecule ligand binding pocket geometries, and many constellations of L-chiral amino acids with a 1.0-Å RMSD to native enzyme active sites. For a representative subset containing 550 active site geometries matching 457 (2) 4-digit (3-digit) enzyme classification (E.C.) numbers, native active site amino acids were generated at random for 472 of 550 cases. This increases to 548 of 550 cases when similar residues are allowed. The most frequently generated sequences correspond to ancient enzymatic functions, e.g., glycolysis, replication, and nucleotide biosynthesis. Surprisingly, even without selection, demi-chiral proteins possess the requisite marginal biochemical function and structure of modern proteins, but were thermodynamically less stable. If demi-chiral proteins were present, they could engage in early metabolism, which created the feedback loop for transcription and cell formation.

FRAGSITE: A Fragment-Based Approach for Virtual Ligand Screening.

To reduce time and cost, virtual ligand screening (VLS) often precedes experimental ligand screening in modern drug discovery. Traditionally, high-resolution structure-based docking approaches rely on experimental structures, while ligand-based approaches need known binders to the target protein and only explore their nearby chemical space. In contrast, our structure-based FINDSITEcomb2.0 approach takes advantage of predicted, low-resolution structures and information from ligands that bind distantly related proteins whose binding sites are similar to the target protein. Using a boosted tree regression machine learning framework, we significantly improved FINDSITEcomb2.0 by integrating ligand fragment scores as encoded by molecular fingerprints with the global ligand similarity scores of FINDSITEcomb2.0. The new approach, FRAGSITE, exploits our observation that ligand fragments, e.g., rings, tend to interact with stereochemically conserved protein subpockets that also occur in evolutionarily unrelated proteins. FRAGSITE was benchmarked on the 102 protein DUD-E set, where any template protein whose sequence identify >30% to the target was excluded. Within the top 100 ranked molecules, FRAGSITE improves VLS precision and recall by 14.3 and 18.5%, respectively, relative to FINDSITEcomb2.0. Moreover, the mean top 1% enrichment factor increases from 25.2 to 30.2. On average, both outperform state-of-the-art deep learning-based methods such as AtomNet. On the more challenging unbiased set LIT-PCBA, FRAGSITE also shows better performance than ligand similarity-based and docking approaches such as two-dimensional ECFP4 and Surflex-Dock v.3066. On a subset of 23 targets from DEKOIS 2.0, FRAGSITE shows much better performance than the boosted tree regression-based, vScreenML scoring function. Experimental testing of FRAGSITE's predictions shows that it has more hits and covers a more diverse region of chemical space than FINDSITEcomb2.0. For the two proteins that were experimentally tested, DHFR, a well-studied protein that catalyzes the conversion of dihydrofolate to tetrahydrofolate, and the kinase ACVR1, FRAGSITE identified new small-molecule nanomolar binders. Interestingly, one new binder of DHFR is a kinase inhibitor predicted to bind in a new subpocket. For ACVR1, FRAGSITE identified new molecules that have diverse scaffolds and estimated nanomolar to micromolar affinities. Thus, FRAGSITE shows significant improvement over prior state-of-the-art ligand virtual screening approaches. A web server is freely available for academic users at http:/sites.gatech.edu/cssb/FRAGSITE.

AlphaFold 2: Why It Works and Its Implications for Understanding the Relationships of Protein Sequence, Structure, and Function.

AlphaFold 2 (AF2) was the star of CASP14, the last biannual structure prediction experiment. Using novel deep learning, AF2 predicted the structures of many difficult protein targets at or near experimental resolution. Here, we present our perspective of why AF2 works and show that it is a very sophisticated fold recognition algorithm that exploits the completeness of the library of single domain PDB structures. It has also learned local side chain packing rearrangements that enable it to refine proteins to high resolution. The benefits and limitations of its ability to predict the structures of many more proteins at or close to atomic detail are discussed.

MEDICASCY: A Machine Learning Approach for Predicting Small-Molecule Drug Side Effects, Indications, Efficacy, and Modes of Action.

To improve the drug discovery yield, a method which is implemented at the beginning of drug discovery that accurately predicts drug side effects, indications, efficacy, and mode of action based solely on the input of the drug's chemical structure is needed. In contrast, extant predictive methods do not comprehensively address these aspects of drug discovery and rely on features derived from extensive, often unavailable experimental information for novel molecules. To address these issues, we developed MEDICASCY, a multilabel-based boosted random forest machine learning method that only requires the small molecule's chemical structure for the drug side effect, indication, efficacy, and probable mode of action target predictions; however, it has comparable or even significantly better performance than existing approaches requiring far more information. In retrospective benchmarking on high confidence predictions, MEDICASCY shows about 78% precision and recall for predicting at least one severe side effect and 72% precision drug efficacy. Experimental validation of MEDICASCY's efficacy predictions on novel molecules shows close to 80% precision for the inhibition of growth in ovarian, breast, and prostate cancer cell lines. Thus, MEDICASCY should improve the success rate for new drug approval. A web service for academic users is available at http://pwp.gatech.edu/cssb/MEDICASCY.

Antimalarial Peptide and Polyketide Natural Products from the Fijian Marine Cyanobacterium Moorea producens.

A new cyclic peptide, kakeromamide B (1), and previously described cytotoxic cyanobacterial natural products ulongamide A (2), lyngbyabellin A (3), 18E-lyngbyaloside C (4), and lyngbyaloside (5) were identified from an antimalarial extract of the Fijian marine cyanobacterium Moorea producens. Compounds 1 and 1 exhibited moderate activity against Plasmodium falciparum blood-stages with EC50 values of 0.89 and 0.99 µM, respectively, whereas 3 was more potent with an EC50 value of 0.15 nM, respectively. Compounds 1, 4, and 5 displayed moderate liver-stage antimalarial activity against P. berghei liver schizonts with EC50 values of 1.1, 0.71, and 0.45 µM, respectively. The threading-based computational method FINDSITEcomb2.0 predicted the binding of 1 and 2 to potentially druggable proteins of Plasmodiumfalciparum, prompting formulation of hypotheses about possible mechanisms of action. Kakeromamide B (1) was predicted to bind to several Plasmodium actin-like proteins and a sortilin protein suggesting possible interference with parasite invasion of host cells. When 1 was tested in a mammalian actin polymerization assay, it stimulated actin polymerization in a dose-dependent manner, suggesting that 1 does, in fact, interact with actin.

Chemical space of Escherichia coli dihydrofolate reductase inhibitors: New approaches for discovering novel drugs for old bugs.

Escherichia coli Dihydrofolate reductase is an important enzyme that is essential for the survival of the Gram-negative microorganism. Inhibitors designed against this enzyme have demonstrated application as antibiotics. However, either because of poor bioavailability of the small-molecules resulting from their inability to cross the double membrane in Gram-negative bacteria or because the microorganism develops resistance to the antibiotics by mutating the DHFR target, discovery of new antibiotics against the enzyme is mandatory to overcome drug-resistance. This review summarizes the field of DHFR inhibition with special focus on recent efforts to effectively interface computational and experimental efforts to discover novel classes of inhibitors that target allosteric and active-sites in drug-resistant variants of EcDHFR.

Repurposed FDA-approved drugs targeting genes influencing aging can extend lifespan and healthspan in rotifers.

Pharmaceutical interventions can slow aging in animals, and have advantages because their dose can be tightly regulated and the timing of the intervention can be closely controlled. They also may complement environmental interventions like caloric restriction by acting additively. A fertile source for therapies slowing aging is FDA approved drugs whose safety has been investigated. Because drugs bind to several protein targets, they cause multiple effects, many of which have not been characterized. It is possible that some of the side effects of drugs prescribed for one therapy may have benefits in retarding aging. We used computationally guided drug screening for prioritizing drug targets to produce a short list of candidate compounds for in vivo testing. We applied the virtual ligand screening approach FINDSITEcomb for screening potential anti-aging protein targets against FDA approved drugs listed in DrugBank. A short list of 31 promising compounds was screened using a multi-tiered approach with rotifers as an animal model of aging. Primary and secondary survival screens and cohort life table experiments identified four drugs capable of extending rotifer lifespan by 8-42%. Exposures to 1 µM erythromycin, 5 µM carglumic acid, 3 µM capecitabine, and 1 µM ivermectin, extended rotifer lifespan without significant effect on reproduction. Some drugs also extended healthspan, as estimated by mitochondria activity and mobility (swimming speed). Our most promising result is that rotifer lifespan was extended by 7-8.9% even when treatment was started in middle age.

FINDSITEcomb2.0: A New Approach for Virtual Ligand Screening of Proteins and Virtual Target Screening of Biomolecules.

Computational approaches for predicting protein-ligand interactions can facilitate drug lead discovery and drug target determination. We have previously developed a threading/structural-based approach, FINDSITEcomb, for the virtual ligand screening of proteins that has been extensively experimentally validated. Even when low resolution predicted protein structures are employed, FINDSITEcomb has the advantage of being faster and more accurate than traditional high-resolution structure-based docking methods. It also overcomes the limitations of traditional QSAR methods that require a known set of seed ligands that bind to the given protein target. Here, we further improve FINDSITEcomb by enhancing its template ligand selection from the PDB/DrugBank/ChEMBL libraries of known protein-ligand interactions by (1) parsing the template proteins and their corresponding binding ligands in the DrugBank and ChEMBL libraries into domains so that the ligands with falsely matched domains to the targets will not be selected as template ligands; (2) applying various thresholds to filter out falsely matched template structures in the structure comparison process and thus their corresponding ligands for template ligand selection. With a sequence identity cutoff of 30% of target to templates and modeled target structures, FINDSITEcomb2.0 is shown to significantly improve upon FINDSITEcomb on the DUD-E benchmark set by increasing the 1% enrichment factor from 16.7 to 22.1, with a p-value of 4.3 × 10-3 by the Student t-test. With an 80% sequence identity cutoff of target to templates for the DUD-E set and modeled target structures, FINDSITEcomb2.0, having a 1% ROC enrichment factor of 52.39, also outperforms state-of-the-art methods that employ machine learning such as a deep convolutional neural network, CNN, with an enrichment of 29.65. Thus, FINDSITEcomb2.0 represents a significant improvement in the state-of-the-art. The FINDSITEcomb2.0 web service is freely available for academic users at http://pwp.gatech.edu/cssb/FINDSITE-COMB-2 .

Effects of confinement on models of intracellular macromolecular dynamics.

The motions of particles in a viscous fluid confined within a spherical cell have been simulated using Brownian and Stokesian dynamics simulations. High volume fractions mimicking the crowded interior of biological cells were used. Importantly, although confinement yields an overall slowdown in motion, the qualitative effects of motion in the interior of the cell can be effectively modeled as if the system were an infinite periodic system. However, we observe layering of particles at the cell wall due to steric interactions in the confined space. Motions of nearby particles are also strongly correlated at the cell wall, and these correlations increase when hydrodynamic interactions are modeled. Further, particles near the cell wall have a tendency to remain near the cell wall. A consequence of these effects is that the mean contact time between particles is longer at the cell wall than in the interior of the cell. These findings identify a specific way that confinement affects the interactions between particles and points to a previously unidentified mechanism that may play a role in signal transduction and other processes near the membrane of biological cells.

DNA Internal Motion Likely Accelerates Protein Target Search in a Packed Nucleoid.

Transcription factors must diffuse through densely packed and coiled DNA to find their binding sites. Using a coarse-grained model of DNA and lac repressor (LacI) in the Escherichia coli nucleoid, simulations were performed to examine how LacI diffuses in such a space. Despite the canonical picture of LacI diffusing rather freely, in reality the DNA is densely packed, is not rigid but highly mobile, and the dynamics of DNA dictates to a great extent the LacI motion. A possibly better picture of unbound LacI motion is that of gated diffusion, where DNA confines LacI in a cage, but LacI can move between cages when hindering DNA strands move out of the way. Three-dimensional diffusion constants for unbound LacI computed from simulations closely match those for unbound LacI in vivo reported in the literature. The internal motions of DNA appear to be governed by strong internal forces arising from being crowded into the small space of the nucleoid. A consequence of the DNA internal motion is that protein target search may be accelerated.

On the importance of composite protein multiple ligand interactions in protein pockets.

Conventional small molecule drug-discovery approaches target protein pockets. However, the limited number of geometrically distinct pockets leads to widespread promiscuity and deleterious side-effects. Here, the idea of COmposite protein LIGands (COLIG) that interact with each other as well as the protein within a single ligand binding pocket is examined. As a practical illustration, experimental evidence that E. coli Dihydrofolate reductase inhibitors are COLIGs is presented. Then, analysis of a non-redundant set of all holo PDB structures indicates that almost 47-76% of proteins (based on different sequence identity thresholds) can simultaneously bind multiple, interacting ligands in the same pocket. Moreover, most ligands that are either Singletons and COLIGs bind at the bottom of ligand binding pocket and occupy 30% and 43% of the volume of the bottom of the pocket. This suggests the use of COLIGs as a potential new class of small molecule drugs. © 2016 Wiley Periodicals, Inc.

A knowledge-based approach for predicting gene-disease associations.

MOTIVATION: Recent advances of next-generation sequence technologies have made it possible to rapidly and inexpensively identify gene variations. Knowing the disease association of these gene variations is important for early intervention to treat deadly diseases and provide possible targets to cure these diseases. Genome-wide association studies (GWAS) have identified many individual genes associated with common diseases. To exploit the large amount of data obtained from GWAS studies and leverage our understanding of common as well as rare diseases, we have developed a knowledge-based approach to predict gene-disease associations. We first derive gene-gene mutual information by utilizing the cooccurrence of genes in known gene-disease association data. Subsequently, the mutual information is combined with known protein-protein interaction networks by a boosted tree regression method. RESULTS: The method called Know-GENE is compared with the method of random walking on the heterogeneous network using the same input data. For a set of 960 diseases, using the same training data in testing in 3-fold cross-validation, the average recall rate within the top ranked 100 genes by Know-GENE is 65.0% compared with 37.9% by the state of the art random walking on heterogeneous network. This significant improvement is mostly due to the inclusion of knowledge-based mutual information. AVAILABILITY AND IMPLEMENTATION: Predictions for genes associated with the 960 diseases are available at http://cssb2.biology.gatech.edu/knowgene CONTACT: : skolnick@gatech.edu.

Pocket detection and interaction-weighted ligand-similarity search yields novel high-affinity binders for Myocilin-OLF, a protein implicated in glaucoma.

Traditional structure and ligand based virtual screening approaches rely on the availability of structural and ligand binding information. To overcome this limitation, hybrid approaches were developed that relied on extraction of ligand binding information from proteins sharing similar folds and hence, evolutionarily relationship. However, they cannot target a chosen pocket in a protein. To address this, a pocket centric virtual ligand screening approach is required. Here, we employ a new, iterative implementation of a pocket and ligand-similarity based approach to virtual ligand screening to predict small molecule binders for the olfactomedin domain of human myocilin implicated in glaucoma. Small-molecule binders of the protein might prevent the aggregation of the protein, commonly seen during glaucoma. First round experimental assessment of the predictions using differential scanning fluorimetry with myoc-OLF yielded 7 hits with a success rate of 12.7%; the best hit had an apparent dissociation constant of 99nM. By matching to the key functional groups of the best ligand that were likely involved in binding, the affinity of the best hit was improved by almost 10,000 fold from the high nanomolar to the low picomolar range. Thus, this study provides preliminary validation of the methodology on a medically important glaucoma associated protein.

Catalytic and substrate promiscuity: distinct multiple chemistries catalysed by the phosphatase domain of receptor protein tyrosine phosphatase.

The presence of latent activities in enzymes is posited to underlie the natural evolution of new catalytic functions. However, the prevalence and extent of such substrate and catalytic ambiguity in evolved enzymes is difficult to address experimentally given the order-of-magnitude difference in the activities for native and, sometimes, promiscuous substrate/s. Further, such latent functions are of special interest when the activities concerned do not fall into the domain of substrate promiscuity. In the present study, we show a special case of such latent enzyme activity by demonstrating the presence of two mechanistically distinct reactions catalysed by the catalytic domain of receptor protein tyrosine phosphatase isoform δ (PTPRδ). The primary catalytic activity involves the hydrolysis of a phosphomonoester bond (C─O─P) with high catalytic efficiency, whereas the secondary activity is the hydrolysis of a glycosidic bond (C─O─C) with poorer catalytic efficiency. This enzyme also displays substrate promiscuity by hydrolysing diester bonds while being highly discriminative for its monoester substrates. To confirm these activities, we also demonstrated their presence on the catalytic domain of protein tyrosine phosphatase Ω (PTPRΩ), a homologue of PTPRδ. Studies on the rate, metal-ion dependence, pH dependence and inhibition of the respective activities showed that they are markedly different. This is the first study that demonstrates a novel sugar hydrolase and diesterase activity for the phosphatase domain (PD) of PTPRδ and PTPRΩ. This work has significant implications for both understanding the evolution of enzymatic activity and the possible physiological role of this new chemistry. Our findings suggest that the genome might harbour a wealth of such alternative latent enzyme activities in the same protein domain that renders our knowledge of metabolic networks incomplete.