3′end cleavage analysis

For the identification of 3′cleavage sites the tool NanoFilt⁵⁰ (Filtering and trimming of long read sequencing data) was used to trim the mRNA sequences from the 5′end to a uniform length so that the last 200 nucleotides (python get_read_ends.py --bases_from_end 200 reads.fastq.gz | gzip > last_200_bp.fastq.gz) are retained. Trimmed reads were aligned to the GRCh38 or GRCh37 human genome references using minimap2. The quantification of 3′ cleavage was performed with Seqmonk using the human polyA database.⁵² Genes that contain more than two annotated polyadenylation sites (PAS) and more than 5 reads at distal PAS sites were selected for downstream analysis. The ratios between the proximal- and distal cleavage sites were calculated.