Cross-Predicting Essential Genes between Two Model Eukaryotic Species Using Machine Learning.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Model organisms like Caenorhabditis elegans and Drosophila melanogaster are crucial for understanding metazoan molecular and cellular processes.
• Functional genomics has identified essential and non-essential genes within species, but cross-species prediction remains challenging.
• Machine learning (ML) approaches have enabled essentiality predictions within species, but not yet between them.

Purpose of the Study:

• To discover and validate common features of essential genes in C. elegans and D. melanogaster using ML.
• To enable cross-species prediction of gene essentiality.
• To identify features that predict gene essentiality across different metazoan species.

Main Methods:

• Utilized machine learning (ML) to analyze features associated with gene essentiality.
• Investigated features including nucleotide/protein sequences, protein orthology, subcellular localization, single-cell RNA-seq, and histone methylation markers.
• Performed complementary analyses on gene functions and chromosomal locations.

Main Results:

• Demonstrated the possibility of cross-species prediction of gene essentiality between C. elegans and D. melanogaster.
• Identified a subset of features (sequence, orthology, localization, scRNA-seq, histone methylation) crucial for cross-species prediction.
• Found essential genes are enriched in transcription and translation functions and located away from heterochromatin.

Conclusions:

• Cross-species prediction of gene essentiality is achievable using specific molecular and genomic features.
• The identified features provide a foundation for predicting essential genes in both model and non-model metazoans.
• This work advances comparative genomics and functional gene essentiality studies.