LDAEXC: LncRNA-Disease Associations Prediction with Deep Autoencoder and XGBoost Classifier.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Long non-coding RNAs (lncRNAs) play crucial roles in human complex diseases.
• Identifying lncRNA-disease associations is vital for disease diagnosis, prognosis, and therapy.
• Traditional experimental methods for identifying these associations are time-consuming and costly.

Purpose of the Study:

• To develop an accurate computational framework, LDAEXC, for inferring lncRNA-disease associations.
• To improve upon existing computational methods for predicting disease-related lncRNAs.

Main Methods:

• LDAEXC integrates multiple similarity views of lncRNAs and diseases for feature construction.
• A deep autoencoder is employed for dimensionality reduction of constructed features.
• An XGBoost classifier is utilized to predict lncRNA-disease associations.

Main Results:

• LDAEXC achieved high Area Under the Curve (AUC) scores in fivefold cross-validation experiments (e.g., 0.9676 ± 0.0043 on one dataset).
• Performance significantly surpassed other advanced computational methods across multiple datasets.
• Case studies on colon and breast cancers demonstrated the framework's practical utility and predictive power.

Conclusions:

• LDAEXC provides an accurate and efficient method for inferring lncRNA-disease associations.
• The framework holds promise for advancing research in complex disease mechanisms and therapeutic strategies.
• LDAEXC's superior performance highlights the potential of deep learning and ensemble methods in bioinformatics.