DeepLncLoc

DeepLncLoc predicts the subcellular localization of long non-coding RNAs (lncRNAs; >200 nucleotides) using a deep learning framework to infer cellular compartmentalization relevant to lncRNA function.

Key Features:

• Subsequence Embedding: Introduces a subsequence embedding method that preserves sequence-order information by dividing lncRNA sequences into consecutive subsequences, extracting patterns from each, and combining them into a comprehensive sequence representation.
• Text Convolutional Neural Network: Applies a text CNN to the embedded sequences to learn high-level sequence features for subcellular localization prediction.
• Performance: Demonstrates superior performance compared to traditional machine learning models based on k-mer features and existing predictors for lncRNA subcellular localization.

Scientific Applications:

• lncRNA functional inference: Infers potential lncRNA functional roles by predicting their subcellular localization.
• Molecular mechanism and disease studies: Supports investigation of molecular mechanisms and disease pathogenesis by linking lncRNA localization to cellular processes.
• Sequence-based prediction: Provides a sequence representation approach applicable to other sequence-based prediction tasks in genomics and proteomics.

Methodology:

The approach divides lncRNA sequences into consecutive subsequences to extract and combine local patterns (subsequence embedding), then uses a text convolutional neural network to learn features from the embedded sequences for localization prediction.