DeepMiRBP: a hybrid model for predicting microRNA-protein interactions based on transfer learning and cosine similarity DOI Creative Commons

Sasan Azizian,

Juan Cui

BMC Bioinformatics, Год журнала: 2024, Номер 25(1)

Опубликована: Дек. 18, 2024

Interactions between microRNAs and RNA-binding proteins are crucial for microRNA-mediated gene regulation sorting. Despite their significance, the molecular mechanisms governing these interactions remain underexplored, apart from sequence motifs identified on microRNAs. To date, only a limited number of microRNA-binding have been confirmed, typically through labor-intensive experimental procedures. Advanced bioinformatics tools urgently needed to facilitate this research. We present DeepMiRBP, novel hybrid deep learning model specifically designed predict by modeling interactions. This innovation approach is first target direct small RNAs proteins. DeepMiRBP consists two main components. The component employs bidirectional long short-term memory (Bi-LSTM) neural networks capture sequential dependencies context within RNA sequences, attention enhance model's focus most relevant features transfer apply knowledge gained large dataset RNA-protein binding sites specific task predicting microRNA-protein Cosine similarity applied assess similarities. second utilizes Convolutional Neural Networks (CNNs) process spatial data inherent in protein structures based Position-Specific Scoring Matrices (PSSM) contact maps generate detailed accurate representations potential achieved prediction accuracy 87.4% during training 85.4% using testing, with an F score 0.860. Additionally, we validated our method three case studies, focusing such as miR-451, -19b, -23a, -21, -223, -let-7d. successfully predicted known miRNA recently discovered proteins, including AGO, YBX1, FXR2, various exosomes. Our proposed strategy represents its kind interaction prediction. Its promising performance underscores uncover critical sorting packaging, well infer new transporter methodologies insights offer scalable template future research, mechanistic discovery disease-related cell-to-cell communication, emphasizing adaptability developing RNA-centric therapeutic interventions personalized medicine.

Язык: Английский

Exosomal insights into ovarian cancer stem cells: revealing the molecular hubs DOI Creative Commons
Kiana Sojoudi, Maryam Solaimani, Hossein Azizi

и другие.

Journal of Ovarian Research, Год журнала: 2025, Номер 18(1)

Опубликована: Янв. 31, 2025

Ovarian cancer is a deadly disease, often diagnosed at advanced stages due to lack of reliable biomarkers. Exosomes, which carry variety molecules such as proteins, lipids, DNA, and non-coding RNAs, have recently emerged promising tools for early detection. While exosomes been studied in various types, comprehensive network analyses exosome proteins ovarian remain limited. In this study, we used protein-protein interaction (PPI) network. Using the Clustermaker2 app MCODE algorithm, identified six significant clusters within network, highlighting regions involved functional pathways. A four-fold algorithmic approach, including MCC, DMNC, Degree, EPC, 12 common hub genes. STRING analysis visualization techniques provided detailed understanding biological processes associated with these Notably, 91.7% genes were translational processes, showing an important role protein synthesis regulation cancer. addition, miRNAs LncRNAs carried by exosomes. These findings highlight potential biomarkers detection therapeutic targets.

Язык: Английский

Процитировано

0
EXTRACELLULAR VESICLES FROM CYCLIC MICE MODULATE LIVER TRANSCRIPTOME IN ESTROUPAUSE MICE INDEPENDENT OF AGE DOI
Bianka M. Zanini, Bianca M. Ávila, Jéssica D. Hense

и другие.

Molecular and Cellular Endocrinology, Год журнала: 2025, Номер 600, С. 112508 - 112508

Опубликована: Фев. 25, 2025

Язык: Английский

Процитировано

0
Non-coding RNAs: emerging biomarkers and therapeutic targets in cancer and inflammatory diseases DOI Creative Commons
Basma Hossam Abdelmonem,

Lereen T. Kamal,

Lilian Waheed Wardy

и другие.

Frontiers in Oncology, Год журнала: 2025, Номер 15

Опубликована: Март 10, 2025

Non-coding RNAs (ncRNAs) have a significant role in gene regulation, especially cancer and inflammatory diseases. ncRNAs, such as microRNA, long non-coding RNAs, circular alter the transcriptional, post-transcriptional, epigenetic expression levels. These molecules act biomarkers possible therapeutic targets because aberrant ncRNA has been directly connected to tumor progression, metastasis, response therapy research. ncRNAs’ interactions with multiple cellular pathways, including MAPK, Wnt, PI3K/AKT/mTOR, impact processes like proliferation, apoptosis, immune responses. The potential of RNA-based therapeutics, anti-microRNA microRNA mimics, restore normal is being actively studied. Additionally, tissue-specific patterns ncRNAs offer unique opportunities for targeted therapy. Specificity, stability, responses are obstacles use ncRNAs; however, novel strategies, modified oligonucleotides delivery systems, developed. profiling may result more individualized successful treatments precision medicine advances, improving patient outcomes creating early diagnosis monitoring opportunities. current review aims investigate roles diseases, focusing on their mechanisms regulation implications non-invasive diagnostics therapies. A comprehensive literature was conducted using PubMed Google Scholar, research published between 2014 2025. Studies were selected based rigorous inclusion criteria, peer-reviewed status relevance Non-English, non-peer-reviewed, inconclusive studies excluded. This approach ensures that findings presented high-quality relevant sources.

Язык: Английский

Процитировано

0
EXOSOMES FROM CYCLIC MICE MODULATE LIVER TRANSCRIPTOME IN ESTROUPAUSE MICE INDEPENDENT OF AGE DOI Creative Commons
Bianka M. Zanini, Bianca M. Ávila, Jéssica D. Hense

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Ноя. 6, 2024

Abstract Background Exosomes are extracellular vesicles secreted by cells that contain microRNAs (miRNAs). These miRNAs can induce changes in gene expression and function of recipient cells. In different exosome content change with age physiological state affecting tissues health. Aims Therefore, the aim this study was to characterize miRNA role exosomes from cyclic female mice modulation liver transcriptome estropausal mice. Main Methods Two-month-old were induced estropause using 4-vinylcyclohexene diepoxide (VCD). At six months VCD-treated divided control group (VCD) treated (VCD+EXO), which received 10 injections at 3-day intervals extracted serum (CTL). Key findings Exosome injection had no effect on body mass, insulin sensitivity or organ weight. We observed ten differentially regulated VCD compared CTL we 931 genes expressed VCD+EXO Interestingly, eight pathways up-regulated treatment down-regulated treatment, indicating reverse promoted liver. Cyp4a12a is male-specific increased females not reversed treatment. Significance Our indicate independent age. Additionally, partially transcriptome.

Язык: Английский

Процитировано

0
DeepMiRBP: a hybrid model for predicting microRNA-protein interactions based on transfer learning and cosine similarity DOI Creative Commons

Sasan Azizian,

Juan Cui

BMC Bioinformatics, Год журнала: 2024, Номер 25(1)

Опубликована: Дек. 18, 2024

Interactions between microRNAs and RNA-binding proteins are crucial for microRNA-mediated gene regulation sorting. Despite their significance, the molecular mechanisms governing these interactions remain underexplored, apart from sequence motifs identified on microRNAs. To date, only a limited number of microRNA-binding have been confirmed, typically through labor-intensive experimental procedures. Advanced bioinformatics tools urgently needed to facilitate this research. We present DeepMiRBP, novel hybrid deep learning model specifically designed predict by modeling interactions. This innovation approach is first target direct small RNAs proteins. DeepMiRBP consists two main components. The component employs bidirectional long short-term memory (Bi-LSTM) neural networks capture sequential dependencies context within RNA sequences, attention enhance model's focus most relevant features transfer apply knowledge gained large dataset RNA-protein binding sites specific task predicting microRNA-protein Cosine similarity applied assess similarities. second utilizes Convolutional Neural Networks (CNNs) process spatial data inherent in protein structures based Position-Specific Scoring Matrices (PSSM) contact maps generate detailed accurate representations potential achieved prediction accuracy 87.4% during training 85.4% using testing, with an F score 0.860. Additionally, we validated our method three case studies, focusing such as miR-451, -19b, -23a, -21, -223, -let-7d. successfully predicted known miRNA recently discovered proteins, including AGO, YBX1, FXR2, various exosomes. Our proposed strategy represents its kind interaction prediction. Its promising performance underscores uncover critical sorting packaging, well infer new transporter methodologies insights offer scalable template future research, mechanistic discovery disease-related cell-to-cell communication, emphasizing adaptability developing RNA-centric therapeutic interventions personalized medicine.

Язык: Английский

Процитировано

0