Related Experiment Video
Updated: May 15, 2026
09:48
Discrimination and Characterization of Heterocellular Populations Using Quantitative Imaging Techniques
Published on: June 30, 2017
7.5K
PScL-DDCFPred: an ensemble deep learning-based approach for characterizing multiclass subcellular localization of
Matee Ullah1, Fazal Hadi1, Jiangning Song2,3
1School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China.
Bioinformatics (Oxford, England)
|June 30, 2022
Summary
We developed PScL-DDCFPred, a novel computational method for predicting protein subcellular localization in human tissues. This approach accurately identifies protein locations, aiding cellular function research and drug design.
Area of Science:
- Bioinformatics and computational biology
- Cellular biology and proteomics
Background:
- Protein subcellular localization is crucial for understanding cellular functions and guiding drug design.
- Accurate prediction of protein localization is a long-standing challenge in bioinformatics.
Purpose of the Study:
- To develop a novel bioimage-based computational approach, PScL-DDCFPred, for accurate prediction of protein subcellular localization in human tissues.
- To improve the accuracy and generalization capability of protein subcellular localization prediction methods.
Main Methods:
- PScL-DDCFPred extracts multiview image features (global and local).
- An integrative feature selection method using stepwise discriminant analysis and generalized discriminant analysis identifies optimal feature sets.
- A classifier combining deep neural network (DNN) and deep-cascade forest (DCF) is established.
Main Results:
- PScL-DDCFPred demonstrated superior performance compared to state-of-the-art methods in 10-fold cross-validation tests.
- An independent test set confirmed the method's generalization capability and superiority.
- Excellent performance is attributed to the DNN-DCF combination, feature complementarity, and optimized feature selection.
Conclusions:
- PScL-DDCFPred offers a highly accurate and reliable method for protein subcellular localization prediction.
- The approach advances computational biology tools for protein function and drug discovery research.
