Pcirc

Pcirc predicts plant circular RNAs (circRNAs) from RNA-seq data using a Random Forest classifier trained on rice circRNA and long non-coding RNA (lncRNA) features to enable accurate circRNA identification for plant functional genomics.

Key Features:

• Machine learning prediction: Uses a Random Forest algorithm trained on rice circRNA and lncRNA datasets.
• Feature extraction: Extracts features including open reading frames, k-mer counts, splicing junction sequence coding, splicing signals, and transposable element–related signals prevalent in plants.
• Performance metrics: Reports accuracy, precision, and F1 scores exceeding 0.99 on test datasets.
• Cross-validation: Employs tenfold cross-validation during model training.
• Cross-species validation: Tested on Arabidopsis thaliana and maize in addition to rice.
• Implementation: Implemented in Python 3.

Scientific Applications:

• Plant circRNA identification: Predicts circRNAs from RNA-seq data to generate species-specific circRNA catalogs.
• Functional genomics: Supports investigation of circRNA-mediated gene regulation and their roles in plant biology.

Methodology:

Extract features from rice circRNA and lncRNA datasets; train a Random Forest classifier with tenfold cross-validation; evaluate using accuracy, precision, and F1 scores; apply the trained model to other plant species for testing and validation.