. 2016 Sep 9;6(11):3507–3515. doi: 10.1534/g3.116.034488

Table 2. Comparison of alignment algorithms for Pool-Seq data: summary across data sets.

Algorithm	Poly.	$F_{ST}$ -sim.	$F_{ST}$ -real	Rank-sum
novoalign(l)	2	3	1	6
novoalign(g)	5	2	2	9
clc4(l)	3	5	8	16
bwa mem^a	12	1	4	17
bwa bwasw^a	7	9	3	19
gsnap^a	1	7	11	19
clc4(g)	4	6	10	20
bwa aln^a	6	8	7	21
segemehl^a	11	4	13	28
bowtie2(l)^a	10	13	5	28
bowtie2(g)^a	13	11	6	30
ngm(g)^a	9	12	9	30
ngm(l)^a	8	10	12	30
mrfast^a	14	14	14	42

Ranks of the algorithm in the previous evaluations are shown: overall suitability (poly: Figure 2), allele frequency differences using simulated data ( $F_{ST}$ -sim.: Figure 2), and allele frequency differences using real data ( $F_{ST}$ -real: Figure 2). Algorithms are sorted according to performance with best the performing algorithm shown at the top (minimizing the rank-sum).

Freely available algorithm.