Genomic regions that are identical or highly similar to one another create ambiguity in alignment to the genome, and ambiguous reads are generally discarded.
Despite the fact that RSEM is capable of fully handling reads that map ambiguously or fall into the gene overlapping regions, it proportionally distributes ambiguous reads according to the number of unique reads in overlapping genes.
All the obtained paired-end reads of 16S rRNA gene PCR amplicons were quality filtered and denoised to remove low quality or ambiguous reads.
The raw reads were filtered to generate clean reads by removal of adaptors, ambiguous reads with unknown nucleotides larger than 10%, and low-quality reads where more than half of the bases' qualities were less than five (Quality score < 5).
After removing the adaptor sequences, ambiguous reads and low-quality reads, a total of 19,714,114 clean reads (approximate 1.77 Gbp) were obtained (GenBank Accession Number: SRR6147052), and were further de novo assembled into 16,603 unigenes with an average length of 1110 bp and an N50 of 1894 bp (Table 1).
These ambiguous reads were eliminated to avoid double counting.
Errors that result from ambiguous reads are currently dealt with in the information processing after the experiment.
After computer filtering to remove ambiguous reads, 831,728 (P) and 1,993,962 (N) sequence reads were obtained corresponding to 60,460 (P) and 113,414 (N) unique reads (18 nt∼30 nt).
Here, the fraction of ambiguous reads increases to 2026%2026%
The key challenge in IE is accurate assignment of ambiguous reads to isoforms.
SWA is applied in the final stages, to reliably map ambiguous reads.
