Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “protein representation learning”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 73 records · Page 4Linked to original sources
PubMed2004

Modeling and analysis of heterogeneous regulation in biological networks.

In this study, we propose a novel model for the representation of biological networks and provide algorithms for learning model parameters from experimental data. Our approach is to build an initial model based on extant biological knowledge and refine it to increase the consistency between model predictions and experimental data. Our model encompasses networks which contain heterogeneous biological entities (mRNA, proteins, metabolites) and aims to capture diverse regulatory circuitry on several levels (metabolism, transcription, translation, post-translation and feedback loops, among them). Algorithmically, the study raises two basic questions: how to use the model for predictions and inference of hidden variables states, and how to extend and rectify model components. We show that these problems are hard in the biologically relevant case where the network contains cycles. We provide a prediction methodology in the presence of cycles and a polynomial time, constant factor approximation for learning the regulation of a single entity. A key feature of our approach is the ability to utilize both high-throughput experimental data, which measure many model entities in a single experiment, as well as specific experimental measurements of few entities or even a single one. In particular, we use together gene expression, growth phenotypes, and proteomics data. We tested our strategy on the lysine biosynthesis pathway in yeast. We constructed a model of more than 150 variables based on an extensive literature survey and evaluated it with diverse experimental data. We used our learning algorithms to propose novel regulatory hypotheses in several cases where the literature-based model was inconsistent with the experiments. We showed that our approach has better accuracy than extant methods of learning regulation.

Algorithms↗
PubMed2002

A Bayesian network model for protein fold and remote homologue recognition.

MOTIVATION: The Bayesian network approach is a framework which combines graphical representation and probability theory, which includes, as a special case, hidden Markov models. Hidden Markov models trained on amino acid sequence or secondary structure data alone have been shown to have potential for addressing the problem of protein fold and superfamily classification. RESULTS: This paper describes a novel implementation of a Bayesian network which simultaneously learns amino acid sequence, secondary structure and residue accessibility for proteins of known three-dimensional structure. An awareness of the errors inherent in predicted secondary structure may be incorporated into the model by means of a confusion matrix. Training and validation data have been derived for a number of protein superfamilies from the Structural Classification of Proteins (SCOP) database. Cross validation results using posterior probability classification demonstrate that the Bayesian network performs better in classifying proteins of known structural superfamily than a hidden Markov model trained on amino acid sequences alone.

Amino Acid Sequence↗
PubMed2004

Disulfide connectivity prediction using recursive neural networks and evolutionary information.

MOTIVATION: We focus on the prediction of disulfide bridges in proteins starting from their amino acid sequence and from the knowledge of the disulfide bonding state of each cysteine. The location of disulfide bridges is a structural feature that conveys important information about the protein main chain conformation and can therefore help towards the solution of the folding problem. Existing approaches based on weighted graph matching algorithms do not take advantage of evolutionary information. Recursive neural networks (RNN), on the other hand, can handle in a natural way complex data structures such as graphs whose vertices are labeled by real vectors, allowing us to incorporate multiple alignment profiles in the graphical representation of disulfide connectivity patterns. RESULTS: The core of the method is the use of machine learning tools to rank alternative disulfide connectivity patterns. We develop an ad-hoc RNN architecture for scoring labeled undirected graphs that represent connectivity patterns. In order to compare our algorithm with previous methods, we report experimental results on the SWISS-PROT 39 dataset. We find that using multiple alignment profiles allows us to obtain significant prediction accuracy improvements, clearly demonstrating the important role played by evolutionary information. AVAILABILITY: The Web interface of the predictor is available at http://neural.dsi.unifi.it/cysteines

Algorithms↗
PubMed2004

Impaired learning-dependent cortical plasticity in Huntington's disease transgenic mice.

Huntington's disease (HD) is a genetically transmitted neurodegenerative disorder. The neuropathology in HD is a selective neuronal cell death in several brain regions including cortex. Although changes in synaptic plasticity were shown within the hippocampus and striatum of HD transgenic mice, there are no studies considering neocortical synaptic plasticity abnormalities in HD. We examined the impact of the HD transgene upon learning-dependent plasticity of cortical representational maps. The effect of associative learning, in which stimulation of a row of vibrissae was paired with appetitive stimulus, upon functional representations of vibrissae in the barrel cortex, was investigated with 2-deoxyglucose brain mapping in presymptomatic R6/1 HD mice. In wild-type mice, cortical representation of the row of vibrissae involved in the training was expanded, while in HD mice the representation of this row was not expanded. The results suggest that presymptomatic R6/1 HD transgenic mice show deficits in plasticity of primary somatosensory cortex.

Animals↗
PubMed2005

Differential regulation of cortical NMDA receptor subunits by sensory learning.

NMDA receptor is an important player in neuronal plasticity, including cortical reorganization. In the adult cerebral cortex, the receptor properties are regulated by relative expression of NR2A and NR2B subunits. We have previously found that 3 days of sensory conditioning, in which stimulation of whiskers was paired with a tail shock, induce NMDA-receptor-dependent expansion of metabolically labeled cortical representations of the stimulated vibrissae. Here, we examined the effect of learning-induced cortical reorganization upon expression of NR2A and NR2B NMDA receptor subunits. An increase in NR2A mRNA expression in the barrel of the "trained" row of vibrissae was observed with in situ hybridization 24 h after sensory conditioning. NR2B mRNA expression level did not change. Protein level of both regulatory subunits and obligatory NR1 subunit were examined in P2 fraction. NR2A protein level was found elevated 1 h and 24 h after the sensory conditioning, but not in controls which received only whisker stimulation, signifying that the change was associated with cortical map reorganization. NR2B protein level was transiently elevated in both trained and stimulated control groups. NR1 protein level did not change. The results show that simple sensory learning induces a change in expression of regulatory NMDA receptor subunits, indicating a potential for receptor channel properties modification.

Animals↗
PubMed2006

Short-term sensory learning does not alter parvalbumin neurons in the barrel cortex of adult mice: a double-labeling study.

We have previously reported that a classical conditioning paradigm involving stimulation of a row of facial vibrissae produced expansion of the cortical representation of the activated vibrissae ("trained row"), this was demonstrated by labeling with 2-deoxyglucose in layer IV of the barrel cortex. We have also shown that functional reorganization of the primary somatosensory cortex is accompanied by an increase in the density of small GABAergic cells and glutamate decarboxylase 67-positive neurons in the hollows of barrels representing the "trained row." GABA neurons of the rat neocortex co-localize with calcium-binding proteins [parvalbumin, carletinin, calbindin D28k] and neuropeptides (vasoactive intestinal polypeptide, somatostatin). In the present study we have examined GABAergic parvalbumin-positive, interneurons in the cortical representation of "trained" facial vibrissae after short-term aversive training, in order to determine whether the observed changes in glutamate decarboxylase 67-positive neurons are accompanied by changes in parvalbumin-positive neurons. Using double immunofluorescent staining, it was found that (i) all parvalbumin-positive neurons in the barrel hollows were glutamate decarboxylase 67-positive, (ii) following aversive training density of glutamate decarboxylase 67-positive neurons in barrel hollows increased significantly compared with controls and (iii) density glutamate decarboxylase 67-positive/parvalbumin-positive neurons in "trained" barrel hollows did not change compared with controls. This study is the first to demonstrate that the density of double-labeled glutamate decarboxylase 67-positive/parvalbumin-positive neurons does not alter during cortical plasticity, thus suggesting that some other population (i.e. parvalbumin negative) of GABAergic interneurons is involved in learning-dependent changes in layer IV of the barrel cortex.

Animals↗
PubMed2004

SILVER helps assign peptides to tandem mass spectra using intensity-based scoring.

Tandem mass spectrometry is commonly used to identify peptides (and thereby proteins) that are present in complex mixtures. Peptide identification from tandem mass spectra is partially automated, but still requires human curation to resolve "borderline" peptide-spectrum matches (PSMs). SILVER is web-based software that assists manual curation of tandem mass spectra, using a recently developed intensity-based machine-learning approach to scoring PSMs, Elias et al. In this method, a large training set of peptide, fragment, and peak-intensity properties for both matched and mismatched PSMs was used to develop a score measuring consistency between each predicted fragment ion of a candidate peptide and its corresponding observed spectral peak intensity. The SILVER interface provides a visual representation of match quality between each candidate fragment ion and the observed spectrum, thereby expediting manual curation of tandem mass spectra. SILVER is available online at http://llama.med.harvard.edu/Software.html.

Amino Acid Sequence↗
PubMed2026

BioNeuralNet: a graph neural network based Multi-Omics network data analysis tool.

SUMMARY: Multi-omics data offer unprecedented insights into complex biological systems, yet their high dimensionality, sparsity, and intricate interactions pose significant analytical challenges. Network-based approaches have advanced multi-omics research by effectively capturing biologically relevant relationships among molecular features (e.g., genes, proteins, metabolites). While these methods are powerful for representing molecular interactions, there remains a need for tools specifically designed to effectively utilize these network representations across diverse downstream analyses. To fulfill this need, we introduce BioNeuralNet, a flexible and modular Python framework tailored for end-to-end network-based multi-omics data analysis. BioNeuralNet leverages Graph Neural Networks (GNNs) to learn biologically meaningful low-dimensional representations from multi-omics networks, converting these complex molecular networks into versatile embeddings. BioNeuralNet supports all major stages of multi-omics network analysis, including several network construction techniques, generation of low-dimensional representations, and a broad range of downstream analytical tasks. Its extensive utilities, including diverse GNN architectures, and compatibility with established Python packages (e.g., scikit-learn, PyTorch, NetworkX), enhance usability and facilitate quick adoption. BioNeuralNet is an open-source, user-friendly, and extensively documented framework designed to support flexible and reproducible multi-omics network analysis in precision medicine. AVAILABILITY AND IMPLEMENTATION: The BioNeuralNet library is available via The Python Package Index (PyPI). Source code, documentation, tutorials, and workflows are hosted at https://bioneuralnet.readthedocs.io. Code archived at https://doi.org/10.5281/zenodo.17503083.

Graph Neural Networks↗
PubMed1998

Cytosolic hippocampal PKC and aging: correlation with discrimination performance.

Adult and aged mice were submitted to a discrimination task in a radial maze (regular trials), and then to probe trials requiring them to form relational representations. Three weeks later, animals were again tested for regular and probe trials. Following another interval of 3 weeks, individual hippocampal cytosolic calcium-dependent and -independent PKC activities were measured. Performance of aged animals was impaired on probe but not regular trials and aged mice had lower hippocampal cytosolic calcium-dependent and -independent PKC activities than adults. Performance on probe trials was specifically correlated with calcium-dependent PKC activity. This suggests a specific relationship between the ability to form relational representations and hippocampal cytosolic calcium-dependent PKC activity.

Aging↗
PubMed2005

Impaired visuospatial recognition memory but normal object novelty detection and relative familiarity judgments in adult mice expressing the APPswe Alzheimer's disease mutation.

The present study examined the effects of a human APPswe mutation on object recognition memory in adult Tg2576 mice. The results showed that 14-month old Tg2576 mice were able to detect object novelty as well as control mice, even with delays of up to 24 hr. In addition, transgenic mice showed a normal recency effect and explored the most recently encountered object significantly less than an object encountered earlier in a trial. However, adult Tg2576 mice showed impairments in detecting a change in the relative positions of an array of familiar objects. The results suggest that the formation of representations involving a combination of object identity and spatial information are particularly sensitive to amyloid pathology in adult APPswe mutant mice.

Alzheimer Disease↗
PubMed1997

Taste and olfactory processing in the brain and its relation to the control of eating.

Taste processing in primates through the nucleus of the solitary tract and the primary taste cortex is shown to represent the identity and intensity of taste inputs. In contrast, in the secondary taste cortex in the orbitofrontal area, single neurons respond to the taste of a food only if hunger is present. The neurons here reflect the reward value of taste. They show sensory-specific satiety. In addition to neurons with best responses to sweet, salt, bitter, and sour, there are separate representations of the taste of protein ("umami") and of astringency (e.g., tannic acid). In the orbitofrontal cortex, olfactory inputs converge onto neurons with taste inputs, forming representations of flavor. Neurons in this region may also respond to the sight of food and to its texture. The olfactory representation for some neurons reflects the taste association of odors; and olfactory sensory-specific satiety is represented in this part of the brain. Rapid learning of visual-to-taste associations is also a feature of the neural processing that occurs in the orbitofrontal cortex.

Animals↗
PubMed2026

CLASPP: A unified model for predicting post-translational modifications.

Post-Translational Modifications (PTMs) are a fundamental mechanism for regulating cellular pathways and increasing the functional diversity of the proteome. Accurately predicting the PTM types that are likely to occur at a given site in the primary sequence is a key challenge in functional proteomics. Existing PTM prediction models predominantly focus on either single PTM types or employ ensemble methods that combine multiple models to predict different PTM types. This fragmentation is largely driven by the vast imbalance in data availability across PTM types, making it difficult to predict multiple PTM types with a single model. To address this limitation, we present the Contrastively Learned Attention-based Stratified PTM Predictor (CLASPP), a unified PTM prediction model. CLASPP addresses imbalance challenges by leveraging unsupervised clustering-based undersampling and a novel contrastive learning framework tailored to PTM data. Additionally, our hierarchical data organization and curation are shown to improve CLASPP's performance by balancing the representation of individual PTM types and provides a standardized dataset to train and validate future model designs. Drawing inspiration from advancements in image and natural language processing, the CLASPP model employs a multi-stage training strategy and a high-quality, curated training dataset to improve PTM prediction performance. To uncover what is learned during the contrastive learning stage, the CLASPP model is shown to distinguish known protein kinase substrate specificity profiles as a form of explainability. Finally, we evaluate the application of CLASPP in predicting PTMs in different model organisms and experimentally validated ubiquitination sites in the understudied DCLK3 kinase. Overall, CLASPP represents a unified model for PTM prediction that addresses key bottlenecks in data imbalance and offers new strategies for biological data curation, thereby improving PTM-type prediction performance across diverse organisms.

Protein Processing, Post-Translational↗
PubMed1993

Hidden Markov models and iterative aligners: study of their equivalence and possibilities.

There are many shared attributes between existing iterative aligners and Hidden Markov Model (HMM). A learning algorithm of HMM called Viterbi is the same as the iteration of DP-matching of iterative aligners. HMM aligners can use the result of an iterative aligner initially, incorporate the similarity score of amino acids, and apply the detailed gap cost systems to improve the matching accuracy. On the other hand, the iterative aligner can inherit the modeling capability of HMM, and provide the better representation of the proteins than motifs. In this paper, we present an overview of several iterative aligners which include the parallel iterative aligner of ICOT and the HMM aligner of Haussler's group. We compare the merits and shortcomings of these aligners. This comparison enables us to formulate a better, more advanced aligner through proper integration of the iterative technique and HMM technique.

Algorithms↗
PubMed1995

Predicting protein folding classes without overly relying on homology.

An important open problem in molecular biology is how to use computational methods to understand the structure and function of proteins given only their primary sequences. We describe and evaluate an original machine-learning approach to classifying protein sequences according to their structural folding class. Our work is novel in several respects: we use a set of protein classes that previously have not been used for classifying primary sequences, and we use a unique set of attributes to represent protein sequences to the learners. We evaluate our approach by measuring its ability to correctly classify proteins that were not in its training set. We compare our input representation to a commonly used input representation--amino acid composition--and show that our approach more accurately classifies proteins that have very limited homology to the sequences on which the systems are trained.

Algorithms↗
PubMed1993

Molecular scene analysis: the integration of direct-methods and artificial-intelligence strategies for solving protein crystal structure.

A knowledge-based approach to crystal structure determination is presented. The approach integrates direct-methods and artificial-intelligence strategies to rephrase the structure determination process as an exercise in scene analysis. A general joint probability distribution framework, which allows the incorporation of isomorphous replacement, anomalous scattering and a priori structural information, forms the basis of the direct-methods strategies. The accumulated knowledge on crystal and molecular structures is exploited through the use of artificial-intelligence strategies, which include techniques of knowledge representation, search and machine learning.

Journal Article↗
PubMed2006

Cross-species annotation of basic leucine zipper factor interactions: Insight into the evolution of closed interaction networks.

Dimeric basic leucine zipper (bZIP) factors constitute one of the most important classes of enhancer-type transcription factors. In vertebrates, bZIP factors are involved in many cellular processes, including cell survival, learning and memory, cancer progression, lipid metabolism, and a variety of developmental processes. These factors have the ability to homodimerize and heterodimerize in a specific and predictable manner, resulting in hundreds of dimers with unique effects on transcription. In recent years, several studies have described dimerization preferences for bZIP factors from different species, including Homo sapiens, Drosophila melanogaster, Arabidopsis thaliana, and Saccharomyces cerevisiae. Here, these findings are summarized as novel, graphical representations of closed, interacting protein networks. These representations combine phylogenetic information, DNA-binding properties, and dimerization preference. Beyond summarizing bZIP dimerization preferences within selected species, we have included annotation for a solitary bZIP factor found in the primitive eukaryote, Giardia lamblia, a possible evolutionary precursor to the complex networks of bZIP factors encoded by other genomes. Finally, we discuss the fundamental similarities and differences between dimerization networks within the context of bZIP factor evolution.

Amino Acid Sequence↗
PubMed2003

Lesions of the Basal forebrain cholinergic system impair task acquisition and abolish cortical plasticity associated with motor skill learning.

The contribution of the basal forebrain cholinergic system in mediating plasticity of cortical sensorimotor representations was examined in the context of normal learning. The effects of specific basal forebrain cholinergic lesions upon cortical reorganization associated with learning a skilled motor task were investigated, addressing, for the first time, the functional consequences of blocking cortical map plasticity. Results demonstrate that disrupting basal forebrain cholinergic function disrupts cortical map reorganization and impairs motor learning. Cholinergic lesions do not impair associative fear learning or overall sensorimotor function. These results support the hypothesis that the basal forebrain cholinergic system may be specifically implicated in forms of learning requiring plasticity of cortical representations.

Acetylcholine↗
PubMed2004

Identification of differentially expressed genes induced in the rat brain by acetyl-L-carnitine as evidenced by suppression subtractive hybridisation.

Acetyl-L-carnitine (ALC) is a molecule widely present in the central nervous system (CNS) formed by the reversible acetylation of carnitine. It acts by stimulating energy metabolism. Reported neurobiological effects of this substance include modulation of brain energy and phospholipid metabolism; cellular macromolecules (including neurotrophic factors and neurohormones); synaptic transmission of multiple neurotransmitters. ALC is of considerable interest for its clinical application in Alzheimer's disease and in the treatment of painful neuropathies. There are experimental data that it affects attention and antagonizes deterioration of ability to learn, improving long-term memory. Moreover, ALC influences nonassociative learning of sensitization type in Hirudo medicinalis. These findings are suggesting that ALC might exert its effects by means of new protein synthesis. ALC or saline solution was injected intraperitoneally each day for 21 days in rats. Poly(A)+ RNAs were isolated from control and treated rat brain. Suppression subtractive hybridisation (SSH) method was applied for the generation of subtracted cDNA libraries and the subsequent identification of differentially expressed transcripts after treatments. The technique generates an equalized representation of differentially expressed genes irrespective of their relative abundance, and it is based on the construction of forward and reverse cDNA libraries that allow the identification of the genes that are regulated or switched off/on after ALC treatment. We identified two modulated genes, the isoform gamma of 14-3-3 protein and a precursor of ATP synthase lipid-binding protein, and one gene switched on by the treatment, the heat shock protein hsp72.

14-3-3 Proteins↗