Fig. 6

Gene and transcript read swapping from error correction. a Circos plot showing reads mapping to different loci after using Cluster/Polish for long inter-read error correction. Each line represents one read and the width of each chromosome bin represents the number of reads (combined thickness of each line). The indented line ends shows FLNC read location and non-indented ends shows read allocation after inter-read error correction. This plot shows 34,637 reads from 4799 genes moving to 2793 genes after Cluster/Polish. The reads are organized by chromosome however swapping occurs within chromosome and between chromosomes. b Circos plot as above but after hybrid inter-read correction with LoRDEC. Each line represents a single read moving from one gene to another with 19,064 reads from 2292 genes moving to 2319 genes after LoRDEC error correction. c The PRAMEF8 gene has coverage from 9 FLNC mapped reads (TAMA Low). Five of these reads were clustered and combined with other reads into one cluster read by Cluster/Polish resulting in a jumbled cluster read mapping to the PRAMEF15 gene (Polish pipeline). This suggests a false negative for PRAMEF8 and false positive for PRAMEF15 in the Polish pipeline due to the use of Cluster/Polish