Skip to main content

Table 3 Concordance between the 1052-amplicon and 35-amplicon panels as a function of input DNA mass and adaptive thresholds

From: Evaluation of an integrated clinical workflow for targeted next-generation sequencing of low-quality tumor DNA using a 51-gene enrichment panel

FFPE sample

Input DNA mass (ng)

Platform concordance

Variant calling with the 1052-Amplicon panel

 

Variant caller threshold

Percent annotated

Ti/Tv ratio

Variants/kb

 

FN

FP

TP

GC > AT

Other

   

1

2000

0 (0)

0 (0)

5 (5)

6 (6)

6 (6)

88% (88%)

2.72 (2.72)

0.89 (0.89)

 

1000

0 (0)

0 (0)

5 (5)

7 (6)

6 (6)

79% (63%)

3.48 (4.74)

0.98 (1.26)

 

500

0 (0)

0 (4)

5 (5)

8.5 (6)

7 (6)

80% (18%)

2.35 (12.72)

0.93 (5.10)

 

250

0 (0)

2 (10)

5 (5)

8.5 (6)

7.5 (6)

82% (11%)

2.18 (17.34)

0.85 (11.19)

2

2000

0 (0)

0 (0)

4 (4)

6 (6)

6 (6)

80% (80%)

2.83 (2.83)

0.88 (0.88)

 

1000

0 (0)

0 (2)

4 (4)

7.5 (6)

6 (6)

84% (56%)

2.67 (4.79)

0.84 (1.32)

 

500

0 (0)

0 (8)

4 (4)

8 (6)

7 (6)

83% (16%)

2.96 (15.03)

0.83 (5.47)

 

250

0 (0)

1 (9)

4 (4)

8 (6)

8 (6)

77% (10%)

3.40 (17.43)

0.84 (10.17)

  1. The indicated mass amounts of two FFPE samples were sequenced after target enrichment with the 1052-amplicon panel. For comparison, 2000 ng DNA from each sample was also sequenced using an alternative 35-amplicon TAS panel. The true variants were defined based on sequencing results from the 35-amplicon panel: false negatives (FN) are variants missed by the 1052-amplicon panel; false positives (FP) are variants not detected by the 35-amplicon panel; and true positives (TP) are variants called by both panels. In general, the variant calling thresholds adaptively increase to adjust to the higher backgrounds of variants detected with the lower input DNA mass amounts. The thresholds are set independently for GC > AT hypotheses versus all other hypotheses; they are based on the log of the variant caller score. The adaptive threshold strategy satisfies multiple criteria spanning all positions of the 1052-amplicon panel with respect to: maintaining a high percent of annotated variants (a surrogate for PPV when the true genotype is unknown); acceptable transition to transversion ratio; and acceptable number of variants called per kb. The parenthetical numbers are the results from maintaining a constant (non-adaptive) threshold. Note that the non-adaptive thresholds remain constant to show the drop in percent annotated and the increased call rate. If the thresholds were held constant at, say, 8, then the call rate would decrease suggesting a drop in sensitivity (data not shown).