|
During the alignment of paired short reads to the reference human genome, it may occur that a read aligns to multiple regions with an identical alignment score. In this scenario, the aligner may assign the read to the wrong region. The other read of that pair, however, will still align to the correct genomic locus. The overall result is that an artefactual fusion is generated. This is indeed a powerful source of artefacts in mRNAseq fusion analyses. To overcome this problem, the Alternative Alignments filter, which is activated by default, scans the alignment data for the presence of alternative alignments. If present, these data are processed, together with the exonic database, to identify the corresponding genes. Then, these data are compared with the alignment(s) and gene(s) of the paired read. If a common gene between the two reads is found, then the data is considered an alternative alignment artefact and thus discarded.
|
