NNLDA

NNLDA predicts potential associations between long non-coding RNAs (lncRNAs) and diseases to aid understanding of disease mechanisms at the molecular level.

Key Features:

• Deep learning architecture: NNLDA employs a deep neural network model to learn patterns linking lncRNAs and diseases.
• Scalability and optimization: Training uses mini-batch stochastic gradient descent to handle large-scale datasets efficiently.
• Benchmark dataset and performance: Evaluated on 205,959 interactions between 19,166 lncRNAs and 529 diseases and reported to outperform existing methods.

Scientific Applications:

• Understanding molecular mechanisms: Predicts lncRNA-disease associations to inform how lncRNA perturbations relate to disease biology.
• Facilitating drug discovery: Identifies candidate lncRNA targets that may guide development of targeted therapies and personalized medicine.
• Prioritizing experimental validation: Filters candidate associations computationally to reduce experimental workload and costs.

Methodology:

NNLDA is implemented in Python 3.5 using TensorFlow 1.12.0 with numpy and pandas and trains a deep neural network with mini-batch stochastic gradient descent using experimental data from the LncRNADisease2.0 database.