Unveiling human origins of replication using deep learning: accurate prediction and comprehensive analysis

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Accurate identification of replication origins (ORIs) is vital for understanding human cell growth and developing cancer therapies.
• Existing methods for ORI prediction have limitations in efficiency and precision.

Purpose of the Study:

• To develop an efficient and precise computational approach for identifying human replication origins (ORIs).
• To improve the accuracy of ORI prediction across different human cell lines.

Main Methods:

• Combined the Z-curve method with a deep learning approach to create the Ori-FinderH tool.
• Utilized 10-fold cross-validation for performance evaluation.
• Employed a genetic algorithm with the Ori-FinderH model to generate artificial ORIs.

Main Results:

• Ori-FinderH achieved a high area under the receiver operating characteristic curve (AUC) of 0.9616 for the K562 cell line.
• A cross-cell-line predictive model further improved performance with an AUC of 0.9706.
• Over 98% of generated artificial sequences contained at least one ORI for Hela, MCF7, and K562 cell lines.

Conclusions:

• Ori-FinderH offers a superior computational method for human ORI identification.
• The approach provides more accurate and comprehensive data for experimental research.
• This advancement has the potential to accelerate progress in cell growth and cancer therapy research.