Nowadays,　the　burgeon　of　machine　learning　has　promoted　its　wide　application　in　various　fields.　In　Bioinformatics,　machine　learning　computational　method　has　become　an　indispensable　part.　Its　efficiency　and　simplicity　break　the　expensive　and　inefficient　barriers　of　previous　experimental　methods.　This　article　briefly　reviews　the　application　and　development　of　machine　learning　in　identifying　and　predicting　key　nucleotide　sites　(such　as　6mA,　5mC,　4mC,　etc.)　in　DNA　or　RNA.　Among　them,　supervised　learning　is　one　of　the　main　methods　to　this　work.　Its　wide　application　has　provided　great　convenience　for　the　development　of　identification　and　prediction　of　key　nucleotides　in　Bioinformatics.　©　2019　IEEE.