Recognition of 3' UTR Stem-Loop Structural Patterns in Alu Transposons in Human Genome

The L1 transposon family and the parasitic one Alu cover more than 45% of the human genome length, which makes them one of the most important DNA sequence components. Recent studies revealed the presence of a highly conserved stem-loop structure located at the 3'-end in both transposon families. It was supposed that this stem-loop is recognized by the ORF2p reverse transcriptase, the reverse transcriptase enzyme encoded by ORF2 segment of L1 transposon for transposition purposes. Moreover, a machine learning-based classifier classifier recognizing these stem-loops by their secondary structure physico-chemical and geometric feature set. In this work a similar classifier is constructed for the target stem-loops at 3'-end of Alu transposons. This is done by conversion of a transposon to a 10-dimensional vector of real numbers which characterize the stem-loop by and its subsequent classification. The prediction accuracy varies between 90% and 94%. The Shift, enthalpy and entropy have the most significant weights in the constructed machine learning algorithm. Keywords: transposons, Alu, L1, stem-loop RNA structures