In recent years, RNA-modifying enzymes have gained significant attention due to their impact on critical RNA-based processes and, consequently, human pathology. However, identifying sites of modifications throughout the transcriptome remains challenging largely due to the lack of accurate and sensitive detection technologies. Recently, we described PhOxi-seq as a method capable of confirming known sites of m2G within abundant classes of RNA, namely, purified rRNA and purified tRNA. Here, we further explore the selectivity of PhOxi-seq and describe an optimized PhOxi-seq workflow, coupled to a novel bioinformatic pipeline, that is capable of detecting enzyme-dependent m2G sites throughout the transcriptome. In this way, we generated a database of potential THUMPD3-dependent m2G sites in multiple RNA classes within a human cancer cell line and further identify potential non-THUMPD3 controlled m2G sites. These potential sites should serve as the basis for further confirmation studies for m2G within the human transcriptome.