Coverage, read length and variant calling accuracy

[ymc]

It seems to me people always think coverage is the more the better. But if such coverage is gained by increasing the read length but with lower quality, can such coverage gain really be converted into better variant calling?

For example, when you go from HiSeq 2500 rapid mode 2x100 to 2x150, you gained 50% coverage but do you actually have more accurate variant calling?

[arm55]

You seem to be confusing coverage and read length. Coverage is the depth of the sequencing, or the average number of reads covering any particular nucleotide.

When you go from 2x100 to 2x150, you're increasing the read length, not necessarily the coverage. There are advantages to having longer reads (notably, it's easier to detect larger variations), however higher coverage does provide a significant bump in detection power.

EDIT: Also, with regards to the HiSeq 2500 rapid mode, the advantage is a reduction in time and cost.

[swbarnes2]

Unless you are looking for variants in a pool, where you expect the variant to only be a tiny fraction of the overall % of reads covering that position, yeah, I think you are better off with more coverage, even if the base quality is a little lower.