Sequencing and Raw Sequence Data Quality Control    ◾    27

Reads with such random scattered hot areas (errors) are hard to be fixed. Another

cause of the loss of quality is the failure of the imaging system at the edges of the flow

cell to read signals (Figure 1.17b). However, the reads in such case can be still useable.

The loss of quality that begins from somewhere and continues until the end of the run

(Figure 1.17c) is usually due to an obstruction to the imaging system (because of a dirt

on the surface of the flow cell). The reads with low quality, in this case, can be filtered

out. A final type of errors associated with tiles is a temporary loss of quality in specific

base positions (Figure 1.17d) due to an obstruction to few cycles (because of bubbles). The

obstruction usually stops the imaging and also prevents the reagents from getting to the

DNA template clusters. Such quality problem may introduce false insertion to the reads

and it cannot be fixed with sequence trimming since the low-quality bases are not at the

end of the reads.

Indeed, not every quality problem associated with tiles can be removed by trimming

or filtering out reads originating from the affected tiles. However, the last one may affect

several tiles and may extend several reads without lowering the overall sequence score. We

should watch out to this kind of fault especially for reads used for variant discovery.

FIGURE 1.17  Per tile sequence quality graphs, some with quality problems.