the end ofND2, at the beginning ofND4andND5, at the end of
ND5, and in control region. The capture sensitivity are 11.49% and
15.65% forRanasp.2 andOnychodactylussp., respectively; the
capture efficiency are 21.32% and 18.25% forRanasp.2 andOny-
chodactylussp., respectively. On the other hand, to evaluate the
capture sensitivity and specificity for LR-LCH, we observed that
there is no gap forOnychodactylussp. (Fig.2b: green) and only one
400 bp gap in the repetitive region forRanasp.2 (Fig.2a: green).
Capture sensitivity increased from an average of 13.57% to an
average of 19.80%; capture efficiency also increased from an average
of 72.56% to 97.71% (by using sites with sequence depth larger
than 10).5 Discussion
Above, we were able to demonstrate that LR-LCH is effective for
capturing divergent loci in the two amphibian metagenomes.
Below, we discuss the following: (1) only using sequence diversity
rate to weigh capture ability of probe is not very accurate; (2) the
advantage and disadvantage of the LR-LCH; and (3) its prospect.
Conserved sequences are similar or identical DNA sequences
across species. In mitogenomes,16s rRNA,12s rRNA,COX1,
COX2, andCOX3are relatively conservative with comparison of
apt6,apt8, NADH dehydrogenase genes, and control region. We
noticed that the conservative parts of mitogenomes were captured
easier than unconservative parts even though the diversity rate of
former is much higher than the latter one. For example,CO1
between Ranasp.1 and Ranasp.2 differed by approximately
25.5% (K2P distance); this variation was greater than the K2P of
15% for the regions betweenND5andND6of the two species,
which is a gap. This indicated that capture ability is not fixed in a
specific threshold for any DNA fragment. We may first consider the
conservativeness of DNA sequence when evaluating and/or
designing probe.
The first advantage is it reduces difficulty of de novo assembly
with on-target reads. We use bait and iterative mapping strategy to
recover these mitogenomes or other small-sized genome regions
(for detailed steps, refer to [25]) and evaluate the performance
according to computational time and iterative number. Although
the number of on-target reads does not increase greatly, the reads
distributed to the genome become even as compared to reads
generated by using the standard capture hybridization. This
reduced iterative number to two to three times and saved compu-
tational time of de novo assembly during each iteration, especially
using MIRA assembler [26].
Another advantage is the length of libraries can be adjusted
according to the sample set of interest. Here in these amphibian40 Xing Chen et al.