Fig. 3

Template switching (TS) overview and layout of read distribution based on read start base. Panel a: Canonical expectation – the reverse transcriptase (RT) transcribes the cap structure as an unencoded cytosine, adds one to two additional cytosines beyond the cap structure, and switches templates. When the 3”G” tail of the TS oligo anneals to these additional cytosines, many of the resulting reads will start with a “G” that does not map to the genome. Some of the unencoded “G”s will artificially appear as encoded if they align to genome-encoded “G”s. Letters colored in red represent bases not found in the genome, letters in blue or black represent encoded nucleotides. An arrow indicates the position of the peak mode, i.e., the most-frequently sequenced starting nucleotide. Panel b: Non-canonical expectation – template switching in which an uncapped RNA starting with a regular “G” is used. Panel c: Diagram of strand invasion that occurs as a result of RNA complementarity to the TS oligo. Panel d: All identified genes were divided into four even quartiles numbered from 5′ - > 3′ position relative to the gene sequence. The number of reads whose 5′-most base resided in a given quartile was counted, and the percent of genes with the majority of reads aligning to that gene quartile was plotted. As expected, more reads mapped toward the 5′ end of genes, particularly reads that began with a “G” either encoded (G), or not encoded in the genome (G’)