BMC Genomics. 2015 Aug 13;16(1):597. doi:10.1186/s12864-015-1670-6

Cumbie JS, Ivanchenko MG, Megraw M.

NanoCAGE-XL and CapFilter: an approach to genome wide identification of high confidence transcription start sites.

Uses the average G-addition per cluster as a threshold. Notes that G-addition is higher when the riboguanosine linker is preceded by AT-rich sequences.