@@ -107,7 +107,7 @@ To illustrate the performances of `ROCK` on a real-case high coverage WGS datase
The two FASTQ files were first processed using `AlienTrimmer` [@Criscuolo:2013] to trim off 5'/3' regions containing many sequencing errors (i.e. Phred score cutoff $Q \! =\!$ 20) or stretches of identical nucleotides.
After discarding too short (i.e. $<\!$ 100 bps) and unpaired HTSRs, a total of 4,973,401 PE HTSRs remained (2,756,008,939 bps, average length $\ell\! =\!$ 277 bps), corresponding to a coverage depth of ~722$\times$.
To reduce this high coverage depth to $c \! = \!$ 50$\times$, `ROCK` (v1.9.5) was run with $k \! =\!$ 25 and $\kappa\! =\! c / \varepsilon\!\approx\!$ 45.
To reduce this high coverage depth to $c \! = \!$ 50$\times$, `ROCK` (v1.9.6) was run with $k \! =\!$ 25 and $\kappa\! =\! c / \varepsilon\!\approx\!$ 45.
To assess the optimal CMS size $\lambda$, the total number $n \! =\!$ 105,584,331 of distinct canonical _k_-mers was estimated using `ntCard` [@Mohamadi:2017], and next specified to `ROCK`, leading to $\lambda\! =\!$ 1.
In theory, using $\kappa\! =\!$ 45 is expected to yield a subset of $cl/(2\ell)\!\approx\!$ 344,000 PE HTSRs totaling $cl \!\approx\!$ 191 Mbps.
Obtained results are summarized in Table \ref{Table 1}.
...
...
@@ -117,15 +117,15 @@ Table: Running times (min:sec), and numbers of PE HTSRs and corresponding base p
For comparison sake, comparable standard digital normalizations ($\kappa\! =\!$ 45 and $k \! =\!$ 25) were also carried out on the same computer (AMD Epyc 2.2 GHz processor, 128 Gb RAM) using other dedicated tools (see Statement of need).