The Local Realignment tool may realign some reads from the same read mapping slightly differently in different runs when run on multithreaded systems. This issue can thus occur if a read mapping used as input has reads with gaps that have been left aligned as well as reads with gaps that have been right aligned. Reads in such mappings that can be affected are identical reads with the same start and end positions with gaps (insertions or deletions) that are differently aligned.
This issue can affect the realignment of read mappings created by the Create UMI Reads tool of the Biomedical Genomics Analysis plugin, where gaps in reads can be left aligned or right aligned. The issue is most likely to affect data that contain repetitive regions or homopolymeric regions, such as Ion Torrent data.
This issue does not affect the realignment of read mappings created by other tools in the CLC Genomics Workbench or CLC Genomics Server, such as Map Reads to Reference, Map Reads to Contigs, RNA-Seq Analysis and Map Bisulfite Reads to Reference. Gaps in mappings generated by these are consistently left aligned.Expected impact
If used for variant detection, the minor differences in realigned read mappings due to this issue can lead to a few very high quality variants having slightly different QUAL values. For example, for one of the mappings, a variant called may have the maximum (capped) QUAL value of 200, while for another, it may have a QUAL value of 159.546 (the second highest possible value) or 156.536 (the third highest possible value). When considered in the context of workflows with downstream filtering steps, it is thus possible that some variants with QUAL values may pass a filter in one run, but not in another run.
Workflows distributed with plugins for affected versions of CLC software filter for QUAL values greater than 20, and thus this issue should not affect results generated using these.
A fix for this issue was implemented in CLC Genomics Workbench 12.0.3 and CLC Genomics Server 11.0.3.
