Search NASA⌕ Search

SEARCH · Search NASA

Results for “protein function prediction”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
NASA NTRS2021

Predicting the Functional State of Protein Kinases Using Interpretable Graph Neural Networks

Kinases are a family of proteins that function as molecular switches, regulating several essential cellular activities such as cell proliferation. Dysfunctional kinases are implicated in several types of cancers and hence they are actively pursued as drug targets. Given the vast number of complex kinase structures that are available in the protein data bank (PDB), there is a necessity to develop methodologies that can identify structurally important moieties of the kinases in an automated fashion, for such techniques can be instrumental in identifying novel drug targets. In this work, we develop a graph neural network (GNN) based deep learning framework for classifying the functionally active and inactive states of a large set of eukaryotic protein kinases, making use of their 3D structure from the PDB. We show that GNN based machine learning models can classify protein states with an accuracy greater than 97%. We further use the GNN models to automatically identify regions of the kinases that are important for its function. For this purpose, Gradient-weighted Class Activation Mapping (Grad-CAM) was implemented on the protein graphs. Remarkably, Grad-CAM consistently identifies the highly conserved DFG motif as the most important part of the protein across the entire kinome, without any prior input. Other regions of the hydrophobic core such as the HRD motif were also identified by the interpretable GNN framework, consistent with the literature. We discuss the significance of each of these regions in detail.

Ashwin Ravichandran↗
DOE OSTI2026

EC-Bench: A Benchmark for Enzyme Commission Number Prediction

Enzymes are proteins that catalyze specific biochemical reactions in cells. Enzyme Commission (EC) numbers are used to annotate enzymes in a four-level hierarchy that classifies enzymes based on the specific chemical reactions they catalyze. Accurate EC number prediction is essential for understanding enzyme functions. Despite the availability of numerous methods for predicting EC numbers from protein sequences, there is no unified framework for evaluating and studying such methods systematically. This gap limits the ability of the community to identify the most effective approaches for enzyme annotation. We introduce EC-Bench, a benchmark for EC number prediction, consisting of 1) an initial representative set of existing methods (including homology-based, deep learning, contrastive learning, and language model methods), 2) existing and novel accuracy and efficiency performance metrics, and 3) selected datasets to allow for comprehensive comparative study. EC-Bench is open-source and provides a framework for researchers to not only compare among existing methods objectively under uniform conditions, but also to introduce and effectively evaluate performance of new methods in a comparative framework. To demonstrate the utility of EC-Bench, we perform extensive experimentation to compare the existing EC number prediction methods and establish their advantages and disadvantages in a variety of prediction tasks, namely “exact EC number prediction”, “EC number completion” and (partial or additional) “EC number recommendation”. We find wide variation in the performance of different methods, but also subtle but potentially useful differences in the performance of different methods across tasks and for different parts of the EC hierarchy.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2023

Large language models generate functional protein sequences across diverse families

Deep-learning language models have shown promise in various biotechnological applications, including protein design and engineering. Here, in this paper, we describe ProGen, a language model that can generate protein sequences with a predictable function across large protein families, akin to generating grammatically and semantically correct natural language sentences on diverse topics. The model was trained on 280 million protein sequences from >19,000 families and is augmented with control tags specifying protein properties. ProGen can be further fine-tuned to curated sequences and tags to improve controllable generation performance of proteins from families with sufficient homologous samples. Artificial proteins fine-tuned to five distinct lysozyme families showed similar catalytic efficiencies as natural lysozymes, with sequence identity to natural proteins as low as 31.4%. ProGen is readily adapted to diverse protein families, as we demonstrate with chorismate mutase and malate dehydrogenase.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2019

The evolution of contact prediction: Evidence that contact selection in statistical contact prediction is changing

Over the last few years, the field of protein structure prediction has been transformed by increasingly-accurate contact prediction software. These methods are based on the detection of coevolutionary relationships between residues from multiple sequence alignments. However, despite speculation, there is little evidence of a link between contact prediction and the physico-chemical interactions which drive amino-acid coevolution. Furthermore, existing protocols predict only a fraction of all protein contacts and it is not clear why some contacts are favoured over others. Using a dataset of 863 protein domains, we assessed the physico-chemical interactions of contacts predicted by CCMpred, MetaPSICOV, and DNCON2, as examples of direct coupling analysis, meta-prediction, and deep learning. We considered correctly-predicted contacts and compared their properties against the protein contacts that were not predicted. Predicted contacts tend to form more bonds than non-predicted contacts, which suggests these contacts may be more important than contacts that were not predicted. Comparing the contacts predicted by each method, we found that metaPSICOV and DNCON2 favour accuracy whereas CCMPred detects contacts with more bonds. This suggests that the push for higher accuracy may lead to a loss of physico-chemically important contacts. These results underscore the connection between protein physico-chemistry and the coevolutionary couplings that can be derived from multiple sequence alignments. This relationship is likely to be relevant to protein structure prediction and functional analysis of protein structure and may be key to understanding their utility for different problems in structural biology. We use publicly-available databases. Our code is available for download at http://opig.stats.ox.ac.uk/.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2018

Large-scale analyses of human microbiomes reveal thousands of small, novel genes and their predicted functions

Abstract Small proteins likely abound in prokaryotes, and may mediate much of the communication that occurs between organisms within a microbiome and their host. Unfortunately, small proteins are traditionally overlooked in biology, in part due to the computational and experimental difficulties in detecting them. To systematically identify novel small proteins, we carried out a large comparative genomics study on 1,773 HMP human-associated metagenomes from four different body sites (mouth, gut, skin and vagina). We describe more than four thousand conserved protein families, the majority of which are novel; ~30% of these protein families are predicted to be secreted or transmembrane. Over 90% of the small protein families have no known domain, and almost half are not represented in reference genomes, emphasizing the incompleteness of knowledge in this space. Our analysis exposes putative novel ‘housekeeping’ small protein families, including a potential novel ribosomally associated protein, as well as ‘mammalian-specific’ or ‘human-specific’ protein families. By analyzing the genomic neighborhood of small genes, we pinpoint a subset of families that are potentially associated with defense against bacteriophage. Finally, we identify families that may be subject to horizontal transfer and are thus potentially involved in adaptation of bacteria to the changing human environment. Our study suggest that small proteins are highly abundant and that those of the human microbiome, in particular, may perform diverse functions that have not been previously reported.

Sberro, Hila↗
DOE OSTI2025

Predicting metal-binding proteins and structures through integration of evolutionary-scale and physics-based modeling

Metals are essential elements in all living organisms, binding to approximately 50% of proteins. They serve to stabilize proteins, catalyze reactions, regulate activities, and fulfill various physiological and pathological functions. While there have been many advancements in determining the structures of protein-metal complexes, numerous metal-binding proteins still need to be identified through computational methods and validated through experiments. Here, to address this need, we have developed the ESMBind workflow, which combines evolutionary scale modeling (ESM) for metal-binding prediction and physics-based protein-metal modeling. Our approach utilizes the ESM-2 and ESM-IF models to predict metal-binding probability at the residue level. In addition, we have designed a metal-placement method and energy minimization technique to generate detailed 3D structures of protein-metal complexes. Our workflow outperforms other models in terms of residue and 3D-level predictions. To demonstrate its effectiveness, we applied the workflow to 142 uncharacterized fungal pathogen proteins and predicted metal-binding proteins involved in fungal infection and virulence.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

The Local Topological Free Energy of the SARS-CoV-2 Spike Protein

The novel coronavirus SARS-CoV-2 infects human cells using a mechanism that involves binding and structural rearrangement of its Spike protein. Understanding protein rearrangement and identifying specific amino acids where mutations affect protein rearrangement has attracted much attention for drug development. In this manuscript, we use a mathematical method to characterize the local topology/geometry of the SARS-CoV-2 Spike protein backbone. Our results show that local conformational changes in the FP, HR1, and CH domains are associated with global conformational changes in the RBD domain. The SARS-CoV-2 variants analyzed in this manuscript (alpha, beta, gamma, delta Mink, G614, N501) show differences in the local conformations of the FP, HR1, and CH domains as well. Finally, most mutations of concern are either in or in the vicinity of high local topological free energy conformations, suggesting that high local topological free energy conformations could be targets for mutations with significant impact of protein function. Namely, the residues 484, 570, 614, 796, and 969, which are present in variants of concern and are targeted as important in protein function, are predicted as such from our model.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

The 1 → 3 massive splitting functions from QCD factorization and SCET

Splitting functions are universal functions describing the collinear dynamics of gauge theories, and as such are crucial ingredients for a wide variety of calculations in perturbative QCD. We present analytic results for the triple collinear splitting functions in QCD with a single massive parton. We derive the splitting functions using two distinct methods; first by expanding the squared matrix elements in the collinear limit, and secondly by using soft-collinear effective theory with massive quarks. We find agreement between these two approaches, providing a strong check of our results. Additionally, we also check all iterated and soft limits of our results, finding agreement with predictions from factorization. Our results provide an important ingredient for higher order perturbative calculations involving massive partons, and for the description of the collinear dynamics of heavy flavor jets.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Design and Characterization of a Transcriptional Repression Toolkit for Plants

Regulation of gene expression is essential for all life. Tools to manipulate the gene expression level have therefore proven to be very valuable in efforts to engineer biological systems. However, there are few well-characterized genetic parts that reduce gene expression in plants, commonly known as transcriptional repressors. We characterized the repression activity of a library consisting of repression motifs from approximately 25% of the members of the largest known family of repressors. Combining sequence information with our trans-regulatory function data, we next generated a library of synthetic transcriptional repression motifs with function predicted in advance. After characterizing our synthetic library, we demonstrated not only that many of our synthetic constructs were functional as repressors but also that our advance predictions of repression strength were better than random guesses. Finally, we assessed the functionality of known transcriptional repression motifs from a wide range of eukaryotes. Our study represents the largest plant repressor motif library experimentally characterized to date, providing unique opportunities for tuning transcription in plants.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2016

Schizophyllum commune has an extensive and functional alternative splicing repertoire

Recent genome-wide studies have demonstrated that fungi possess the machinery to alternatively splice pre-mRNA. However, there has not been a systematic categorization of the functional impact of alternative splicing in a fungus. We investigate alternative splicing and its functional consequences in the model mushroom forming fungus Schizophyllum commune. Alternative splicing was demonstrated for 2,285 out of 12,988 expressed genes, resulting in 20% additional transcripts. Intron retentions were the most common alternative splicing events, accounting for 33% of all splicing events, and 43% of the events in coding regions. On the other hand, exon skipping events were rare in coding regions (1%) but enriched in UTRs where they accounted for 57% of the events. Specific functional groups, including transcription factors, contained alternatively spliced genes. Alternatively spliced transcripts were regulated differently throughout development in 19% of the 2,285 alternatively spliced genes. Notably, 69% of alternatively spliced genes have predicted alternative functionality by loss or gain of functional domains, or by acquiring alternative subcellular locations. S. commune exhibits more alternative splicing than any other studied fungus. Finally, taken together, alternative splicing increases the complexity of the S. commune proteome considerably and provides it with a rich repertoire of alternative functionality that is exploited dynamically.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

Allosteric prediction via convolutional neural networks and protein structural and dynamical features

Allostery is the phenomenon whereby a binding event or covalent modification at one site in a protein modulates function at a distal site, thus changing a protein’s functional state. As such, it is a ubiquitous aspect of protein functional regulation. Computationally predicting allosteric states is important as part of the broader challenge of functional annotation, but it also has practical implications for drug development, as targeting an allosteric site often affords greater specificity compared with targeting an orthosteric site. This study introduces a machine learning approach to predict the allosteric functional state using the small G-protein KRas as the model system, due to its implication in many types of cancer and being well studied as a result with many x-ray crystallographic structures of KRas available with different mutations and ligands bound. Using structural and dynamical features that can be cast as images, namely interatomic distances, contact maps, covariance, and mutual information, supervised learning was performed using convolutional neural networks. Two pretrained convolutional neural network architectures, GoogLeNet and ResNet18, were fine-tuned to classify KRas into active or inactive states based on these features. Across training regimes, atomic contact maps emerged as the most effective structural feature, whereas linearized mutual information outperformed covariance in capturing dynamical correlations relevant to allostery. Models achieved significant validation accuracy, with atomic contact maps yielding up to 90% accuracy. In conclusion, the findings suggest that integrating global structural rearrangements and correlated motion patterns with deep learning can reliably predict protein allosteric states, offering a promising framework for understanding allosteric regulation and developing targeted therapeutics.

Rajeshwar T., Rajitha [Oak Ridge National Laborato↗
DOE OSTI2022

Multi-head attention-based U-Nets for predicting protein domain boundaries using 1D sequence features and 2D distance maps

Abstract The information about the domain architecture of proteins is useful for studying protein structure and function. However, accurate prediction of protein domain boundaries (i.e., sequence regions separating two domains) from sequence remains a significant challenge. In this work, we develop a deep learning method based on multi-head U-Nets (called DistDom) to predict protein domain boundaries utilizing 1D sequence features and predicted 2D inter-residue distance map as input. The 1D features contain the evolutionary and physicochemical information of protein sequences, whereas the 2D distance map includes the structural information of proteins that was rarely used in domain boundary prediction before. The 1D and 2D features are processed by the 1D and 2D U-Nets respectively to generate hidden features. The hidden features are then used by the multi-head attention to predict the probability of each residue of a protein being in a domain boundary, leveraging both local and global information in the features. The residue-level domain boundary predictions can be used to classify proteins as single-domain or multi-domain proteins. It classifies the CASP14 single-domain and multi-domain targets at the accuracy of 75.9%, 13.28% more accurate than the state-of-the-art method. Tested on the CASP14 multi-domain protein targets with expert annotated domain boundaries, the average per-target F1 measure score of the domain boundary prediction by DistDom is 0.263, 29.56% higher than the state-of-the-art method.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

Beneath the surface: Unsolved questions in soil virus ecology

Soil virus ecology is an exciting but still nascent field of research in soil microbiology. While there has been a recent surge in soil virus research studies, many fundamental questions remain unanswered, and a range of technical and bioinformatic challenges need to be overcome. In this perspective article, we present a series of key questions that highlight fruitful research areas for ongoing and future efforts. These include describing the challenges involved in understanding soil viral abundance and activity, spatiotemporal dynamics, life strategy prevalence, virus-mediated biogeochemical impacts, viral protein function, host prediction, and soil RNA virus discovery. In the near term, combining approaches (e.g., cultivation-based, meta-omics, biogeochemical, experimental, and bioinformatic) will be key to assessing the ecological and biogeochemical impacts of soil viruses from the microscopic to the field and global scales. Still, we stress that results must be tempered by current methodological limitations and highlight knowledge gaps that are most pressing to fill via new methods or measurements, such as the prevalence of different viral replication strategies across soils, the fate of microbial necromass carbon after viral lysis, the frequency of virus-host encounters that do not lead to successful infections yet could be bioinformatically mistaken as infections, and the diversity and ecological impacts of RNA viruses in soil.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2018

Characterization of Indole-3-acetic Acid Biosynthesis and the Effects of This Phytohormone on the Proteome of the Plant-Associated Microbe Pantoea sp. YR343

Here, indole-3-acetic acid (IAA) plays a central role in plant growth and development, and many plant-associated microbes produce IAA using tryptophan as the precursor. Using genomic analyses, we predicted that Pantoea sp. YR343, a microbe isolated from Populus deltoides , synthesizes IAA using the indole-3-pyruvate (IPA) pathway. To better understand IAA biosynthesis and the effects of IAA exposure on cell physiology, we characterized proteomes of Pantoea sp. YR343 grown in the presence of tryptophan or IAA. Exposure to IAA resulted in upregulation of proteins predicted to function in carbohydrate and amino acid transport and exopolysaccharide (EPS) biosynthesis. Metabolite profiles of wild-type cells showed the production of IPA, IAA, and tryptophol, consistent with an active IPA pathway. Finally, we constructed an ΔipdC mutant that showed the elimination of tryptophol, consistent with a loss of IpdC activity, but was still able to produce IAA (20% of wild-type levels). Although we failed to detect intermediates from other known IAA biosynthetic pathways, this result suggests the possibility of an alternate pathway or the production of IAA by a nonenzymatic route in Pantoea sp. YR343. The Δ ipdC mutant was able to efficiently colonize poplar, suggesting that an active IPA pathway is not required for plant association.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2016

New CRISPR–Cas systems from uncultivated microbes

We present that CRISPR-Cas systems provide microbes with adaptive immunity by employing short DNA sequences, termed spacers, that guide Cas proteins to cleave foreign DNA. Class 2 CRISPR-Cas systems are streamlined versions, in which a single RNA-bound Cas protein recognizes and cleaves target sequences. The programmable nature of these minimal systems has enabled researchers to repurpose them into a versatile technology that is broadly revolutionizing biological and clinical research. However, current CRISPR-Cas technologies are based solely on systems from isolated bacteria, leaving the vast majority of enzymes from organisms that have not been cultured untapped. Metagenomics, the sequencing of DNA extracted directly from natural microbial communities, provides access to the genetic material of a huge array of uncultivated organisms. Here, using genome-resolved metagenomics, we identify a number of CRISPR-Cas systems, including the first reported Cas9 in the archaeal domain of life, to our knowledge. This divergent Cas9 protein was found in little-studied nanoarchaea as part of an active CRISPR-Cas system. In bacteria, we discovered two previously unknown systems, CRISPR-CasX and CRISPR-CasY, which are among the most compact systems yet discovered. Notably, all required functional components were identified by metagenomics, enabling validation of robust in vivo RNA-guided DNA interference activity in Escherichia coli. Lastly, interrogation of environmental microbial communities combined with in vivo experiments allows us to access an unprecedented diversity of genomes, the content of which will expand the repertoire of microbe-based biotechnologies.

59 BASIC BIOLOGICAL SCIENCES↗