A hybrid 1DCNN-GRU deep learning framework for classifying caprine granulosa cell fertility potential using...

Area of Science:

• Reproductive biology
• Bioinformatics
• Genomics

Background:

• Granulosa cells (GCs) are vital for goat follicular development and oocyte quality.
• GC gene expression profiles can indicate fertility, but standardized assessment methods are lacking.
• Transcriptomic data offers potential biomarkers for fertility, yet requires robust analytical tools.

Purpose of the Study:

• To develop a hybrid deep learning model for classifying goat GCs based on fertility potential.
• To utilize single-cell RNA sequencing (scRNA-seq) data for GC classification.
• To create a quantifiable method for assessing GC quality using gene expression.

Main Methods:

• Analysis of publicly available goat scRNA-seq datasets.
• Identification of 44 differentially expressed genes (DEGs) to distinguish fertility-supporting (FS) and non-fertility-supporting (NFS) GCs.
• Training and evaluation of a hybrid 1DCNN-GRU deep learning model using DEG expression profiles.

Main Results:

• The hybrid 1DCNN-GRU model achieved high classification performance (accuracy 98.89%, F1 score 98.84%).
• The model identified a significantly higher proportion of FS-GCs in polytocous goats (87%) compared to monotocous goats (10.17%).
• DEG analysis confirmed the model's biological consistency and generalizability across datasets.

Conclusions:

• This study introduces the first deep learning-based classification of goat GCs using scRNA-seq.
• The developed 1DCNN-GRU model provides a robust and quantifiable method for evaluating GC fertility.
• This approach holds promise for enhancing reproductive selection and precision livestock management.