Validation no abra/ no soft-clipping

[KilianIlius]

Check the relevance of InDel realignment and sof-clipping.

• Exon germline (NA12878x3_26)

  • mapping with bwa
    • no InDel realignment, no soft-clipping
    • no InDel realignment, but soft-clipping
    • InDel realignment, no soft-clipping
    • InDel realignment and soft-clipping
  • Call variants with
    • freebayes
    • DeepVariant
  • validate calls with validate_NA12878.php

• Exon somatic (NA12878x2_79 (as tumor); NA12877_20 (as normal))

  • Construct tumor sample by mixing NA12878x2_79 with NA12877_20
  • tumor-normal analysis
    • mapping with bwa
      • no InDel realignment, no soft-clipping
      • no InDel realignment, but soft-clipping
      • InDel realignment, no soft-clipping
      • InDel realignment and soft-clipping
    • Call variants with
      • strelka
      • dragen
    • validate calls with validate_somatic.php
      • strelka
      • dragen
    • re-validate with 180mio reads
  • tumor-only analysis
    • use "tumor" sample created for tumor-normal analysis
    • Call variants with VarScan2
    • write script to validate tumor-olny calls
    • validate calls

[KilianIlius]

Exon Germline

FreeBayes

Settings	SNV Sensitivity	SNV PPV	SNV F1	SNV Genotyping	InDel Sensitivity	InDel PPV	InDel F1	InDel Genotyping	All Sensitivity	All PPV	All F1	All Genotyping
no_abra no_sc	99.17%	98.30%	98.73%	99.72%	95.08%	95.16%	95.12%	94.87%	98.93%	98.52%	98.12%	99.45%
no_abra with_sc	99.22%	98.39%	98.80%	99.74%	94.28%	95.63%	94.95%	92.30%	98.94%	98.58%	98.23%	99.32%
with_abra no_sc	99.16%	98.32%	98.74%	99.71%	97.07%	93.98%	95.50%	96.70%	99.04%	98.55%	98.06%	99.54%
with_abra with_sc	99.22%	98.40%	98.81%	99.73%	96.74%	94.59%	95.65%	94.94%	99.08%	98.62%	98.17%	99.46%

DeepVariant

Settings	SNV Sensitivity	SNV PPV	SNV F1	SNV Genotyping	InDel Sensitivity	InDel PPV	InDel F1	InDel Genotyping	All Sensitivity	All PPV	All F1	All Genotyping
no_abra no_sc	99.35%	99.83%	99.59%	99.93%	97.47%	99.54%	98.49%	99.72%	99.24%	99.53%	99.81%	99.92%
no_abra with_sc	99.35%	99.83%	99.59%	99.92%	96.69%	99.49%	98.07%	99.48%	99.20%	99.50%	99.81%	99.90%
with_abra no_sc	99.28%	99.82%	99.55%	99.93%	97.75%	98.86%	98.30%	99.86%	99.19%	99.48%	99.77%	99.92%
with_abra with_sc	99.27%	99.83%	99.55%	99.93%	97.23%	98.67%	97.95%	99.63%	99.15%	99.45%	99.76%	99.91%

[KilianIlius]

Exon Somatic

Sensitivity and PPV are compared to an analysis without Softclipping and without Indel Realignment by abra2

Softclipping Dragen vs. Strelka

AF	Type	Recall/Sensitivity (Dragen)	Precision/PPV (Dragen)	Recall/Sensitivity (Strelka)	Precision/PPV (Strelka)
0.05	all	1.68%	-0.07%	-2.65%	-0.03%
0.05	SNVs	1.69%	-0.05%	-2.92%	-0.03%
0.05	InDels	0.53%	-0.88%	-0.48%	0.00%
0.1	all	0.99%	-0.03%	-0.55%	0.06%
0.1	SNVs	0.83%	-0.02%	-0.71%	0.06%
0.1	InDels	1.70%	-0.38%	-0.31%	0.00%
0.2	all	0.33%	-0.04%	0.15%	0.03%
0.2	SNVs	0.12%	-0.03%	0.00%	0.03%
0.2	InDels	2.05%	-0.32%	0.77%	-0.01%
0.4	all	0.24%	-0.06%	-0.06%	0.09%
0.4	SNVs	0.10%	-0.06%	-0.24%	0.08%
0.4	InDels	1.58%	-0.17%	1.34%	0.15%

---

Abra Dragen vs. Strelka

AF	Type	Recall/Sensitivity (Dragen)	Precision/PPV (Dragen)	Recall/Sensitivity (Strelka)	Precision/PPV (Strelka)
0.05	all	0.01%	0.16%	-0.01%	0.00%
0.05	SNVs	0.04%	0.07%	0.01%	0.00%
0.05	InDels	-0.12%	2.60%	-0.06%	0.00%
0.1	all	-0.08%	0.08%	-0.05%	-0.03%
0.1	SNVs	-0.04%	0.04%	-0.01%	-0.01%
0.1	InDels	-0.47%	1.01%	-0.33%	-0.74%
0.2	all	-0.09%	0.12%	-0.02%	-0.01%
0.2	SNVs	-0.01%	0.04%	0.01%	0.00%
0.2	InDels	-0.85%	1.42%	-0.22%	-0.31%
0.4	all	-0.10%	0.05%	-0.08%	-0.02%
0.4	SNVs	-0.03%	0.00%	-0.03%	0.00%
0.4	InDels	-0.93%	0.73%	-0.55%	-0.43%

---

Abra and Softclipping Dragen vs. Strelka

AF	Type	Recall/Sensitivity (Dragen)	Precision/PPV (Dragen)	Recall/Sensitivity (Strelka)	Precision/PPV (Strelka)
0.05	all	1.70%	0.12%	-2.63%	-0.02%
0.05	SNVs	1.70%	0.03%	-2.92%	-0.02%
0.05	InDels	0.72%	2.50%	-0.17%	0.00%
0.1	all	0.89%	0.07%	-0.56%	0.03%
0.1	SNVs	0.77%	0.03%	-0.71%	0.05%
0.1	InDels	1.21%	0.81%	-0.25%	-0.81%
0.2	all	0.28%	0.07%	0.16%	0.02%
0.2	SNVs	0.09%	0.00%	-0.01%	0.04%
0.2	InDels	1.75%	1.19%	1.07%	-0.34%
0.4	all	0.18%	0.01%	-0.12%	0.06%
0.4	SNVs	0.06%	-0.03%	-0.28%	0.08%
0.4	InDels	1.23%	0.62%	1.03%	-0.31%