Translation Initiation Site Prediction based on the TITER and Machine Learning Method

Translation initiation is a crucial step in the regulation of gene expression. The selection ofthe translation initiation site (TIS) is critical, as it establishes the correct open reading frame for mRNAdecoding. Annotated translation initiation sites initiate the translation process and may commence fromseveral alternative TISs, including AUG and non-AUG codons, posing a significant challenge.Nucleotides flanking the repeat region, particularly those in close proximity to the start site, are believedto enhance translation initiation. Consequently, extensive research has been conducted on this issue. Inthis paper, we propose a machine learning-based method to aid in the identification of ATGs andtranslation start sites that are nearly identical. Kozak, random forest classification (RFC), andTranslation Initiation siTE detector (TITER) similarity score algorithms are employed in this article,with the RFC algorithm yielding the most favorable results.The final method is evaluated on ATG RFC and Near-Cognate RFC datasets. The RFC modeldemonstrates an accuracy of ۸۷.۷۹% in ۳۴۴ balanced cases. The experimental results confirm that theproposed approach, in comparison to similar methods, achieves a more concise set of features whilemaintaining high accuracy.

کلیدواژه ها:

codon ، Nucleotide ، Translation mRNA ، machine learning ، TITER