Table 3

Overall performance estimates based on synthetic template simulations

TPR

FPR

FDR


UNI

0.648

2.475e-5

0.028

BESTNO

0.751

2.213e-4

0.176

BESTNO-Q

0.673

1.939e-5

0.044

BEST

0.816

1.106e-3

0.400

ALL

0.744

3.679e-5

0.027


Sniper performance estimates reflect averages taken across a range of mismatches (1 ≤ k ≤ 3) and coverage levels (10×, 25×, 32×, 50×, 75×, 100×, 150×) over five independently sampled replicates and four synthetic reference templates (Table 2) at our recommended stringency cutoff of Q ≥ 40. Note, these performance estimates are conservative because the majority of genotyped loci come from non-unique templates. See Additional file 14 for estimates broken down by template. See Figure 1b for complete descriptions of mapping strategies. See Materials and methods for corresponding equations. ALL, unique and multiply mapped reads; BEST, best guess; BESTNO, best no-guess mapping; BESTNO-Q, same as BESTNO but uses read quality values for mapping; FDR, false discovery rate; FPR, false positive rate; TPR, true positive rate; UNI, strictly unique mapping.

Simola and Kim Genome Biology 2011 12:R55   doi:10.1186/gb-2011-12-6-r55

Open Data