Multi-Label Fusion Collaborative Matrix Factorization for Predicting LncRNA-Disease Associations.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Human diseases are increasingly linked to long non-coding RNAs (lncRNAs).
• Identifying these associations is crucial but time-consuming.
• Efficient prediction methods for lncRNA-disease associations (LDAs) are needed.

Purpose of the Study:

• To develop an efficient computational model for predicting lncRNA-disease associations (LDAs).
• To enhance the accuracy and comprehensiveness of LDA predictions.

Main Methods:

• Proposed a multi-label fusion collaborative matrix factorization (MLFCMF) approach.
• Optimized lncRNA and disease spaces using multi-label learning.
• Integrated Gaussian interaction profile (GIP) kernel for network similarity.
• Employed collaborative matrix factorization for final predictions.

Main Results:

• The MLFCMF method achieved an Area Under the Curve (AUC) of 0.8612 via ten-fold cross-validation.
• Demonstrated effectiveness in predicting associations for ovarian, colorectal, and lung cancers.
• The model proved to be an effective tool for predicting LDAs.

Conclusions:

• MLFCMF is an effective computational model for predicting lncRNA-disease associations.
• The multi-label fusion and collaborative matrix factorization approach improves prediction accuracy.
• This method offers a valuable tool for advancing research in lncRNA and disease.