Prediction of Ligand–Receptor Interactions Based on CatBoost and Deep Forest and Their Application in Cell–Cell Communication Analysis
Uložené v:
| Názov: | Prediction of Ligand–Receptor Interactions Based on CatBoost and Deep Forest and Their Application in Cell–Cell Communication Analysis |
|---|---|
| Autori: | Wei Wu, Zhao Wang, Longlong Liu, Junfeng Huang, Haifan Qiu, Lihong Peng, Libo Nie |
| Rok vydania: | 2025 |
| Zbierka: | The University of Auckland: Figshare |
| Predmety: | Biophysics, Cell Biology, Pharmacology, Cancer, Biological Sciences not elsewhere classified, Information Systems not elsewhere classified, vector using pyfeat, u >< b, point evaluation strategy, merging known interactions, https :// github, freely available tool, mediated cellular communication, intracellular communication networks, potential drug targets, interacting lrps mif, decoded cellular crosstalk, cellcdmt visualized intercellular, cellcdmt filters interactions, unlabeled lrp based, interpreting lri candidates, cell communication, lrps ), drug design, visualizes crosstalk, ular crosstalk, cellcdmt depicts, vital mediators, tfrc may, sigmoid plot |
| Popis: | Cell-to-cell communication (CCC) is prominent for cell growth and development as well as tissue and organ formation. CCC inference can help us to deeply understand cellular interplay and discover potential therapeutic targets for complex diseases. Cells communicate through direct contact or indirect dialogue using interacting ligand–receptor pairs (LRPs). Consequently, the CCC inference generally contains ligand–receptor interaction (LRI) data curation and LRI-mediated communication strength quantification. Here, we introduce a computational method, CellCDmT, to elucidate Cell ular crosstalk. For interpreting LRI candidates, CellCDmT depicts each LRP as a vector using PyFeat, selects their informative features through XGBoost, and classifies each unlabeled LRP based on an ensemble model with C atBoost and D eep forest. For deciphering LRI-mediated cellular communication, CellCDmT filters interactions after merging known interactions and predictions, quantifies communication strength using a T hree-point evaluation strategy with m aximum difference, and visualizes crosstalk through the heatmap view, network view, circos view, and sigmoid plot. Using 8 evaluation metrics, CellCDmT was benchmarked with 7 LRI prediction baselines, 5 state-of-the-art LRI validation tools, and 8 CCC inference competitors. The outcomes demonstrated that CellCDmT accurately classified unlabeled LRPs and decoded cellular crosstalk. Moreover, CellCDmT visualized intercellular and intracellular communication networks in breast cancer. Interacting LRPs MIF-CD74, WNT7B-FZD1, and B2M-TFRC may be vital mediators of breast cancer. Ligands FGF22, B2M, and RSPO4 may be potential drug targets of breast cancer. CellCDmT will be conducive to facilitating our understanding about disease mechanisms and further promoting tumor targeted therapy and drug design. As a freely available tool, CellCDmT can be accessed at https://github.com/plhhnu/CellCDmT. |
| Druh dokumentu: | article in journal/newspaper |
| Jazyk: | unknown |
| Relation: | https://figshare.com/articles/journal_contribution/Prediction_of_Ligand_Receptor_Interactions_Based_on_CatBoost_and_Deep_Forest_and_Their_Application_in_Cell_Cell_Communication_Analysis/29196646 |
| DOI: | 10.1021/acs.jcim.5c00726.s001 |
| Dostupnosť: | https://doi.org/10.1021/acs.jcim.5c00726.s001 https://figshare.com/articles/journal_contribution/Prediction_of_Ligand_Receptor_Interactions_Based_on_CatBoost_and_Deep_Forest_and_Their_Application_in_Cell_Cell_Communication_Analysis/29196646 |
| Rights: | CC BY-NC 4.0 |
| Prístupové číslo: | edsbas.CAB6C062 |
| Databáza: | BASE |
Nájsť tento článok vo Web of Science | Abstrakt: | Cell-to-cell communication (CCC) is prominent for cell growth and development as well as tissue and organ formation. CCC inference can help us to deeply understand cellular interplay and discover potential therapeutic targets for complex diseases. Cells communicate through direct contact or indirect dialogue using interacting ligand–receptor pairs (LRPs). Consequently, the CCC inference generally contains ligand–receptor interaction (LRI) data curation and LRI-mediated communication strength quantification. Here, we introduce a computational method, CellCDmT, to elucidate Cell ular crosstalk. For interpreting LRI candidates, CellCDmT depicts each LRP as a vector using PyFeat, selects their informative features through XGBoost, and classifies each unlabeled LRP based on an ensemble model with C atBoost and D eep forest. For deciphering LRI-mediated cellular communication, CellCDmT filters interactions after merging known interactions and predictions, quantifies communication strength using a T hree-point evaluation strategy with m aximum difference, and visualizes crosstalk through the heatmap view, network view, circos view, and sigmoid plot. Using 8 evaluation metrics, CellCDmT was benchmarked with 7 LRI prediction baselines, 5 state-of-the-art LRI validation tools, and 8 CCC inference competitors. The outcomes demonstrated that CellCDmT accurately classified unlabeled LRPs and decoded cellular crosstalk. Moreover, CellCDmT visualized intercellular and intracellular communication networks in breast cancer. Interacting LRPs MIF-CD74, WNT7B-FZD1, and B2M-TFRC may be vital mediators of breast cancer. Ligands FGF22, B2M, and RSPO4 may be potential drug targets of breast cancer. CellCDmT will be conducive to facilitating our understanding about disease mechanisms and further promoting tumor targeted therapy and drug design. As a freely available tool, CellCDmT can be accessed at https://github.com/plhhnu/CellCDmT. |
|---|---|
| DOI: | 10.1021/acs.jcim.5c00726.s001 |