How to select meaningful CNVs? #46
Description
Hey,
After many pains I've successfully run DECoN on 208 WES samples. For each sample I've got like 300-1000 CNV. Most of them are them are rubbish. Is there any hint how should I filter the last cnv.calls_plot DF in order to refine the results?
ID sample correlation N.comp start.p end.p type nexons start end chromosome id BF reads.expected reads.observed reads.ratio Gene start.b end.b chr
1 1 s1 0.9927666 13 1 1 deletion 1 69037 70008 1 chr1:69037-70008 2.66 309 216 0.699 OR4F5 1 1 chr1
2 2 s1 0.9927666 13 380 380 duplication 1 1645126 1645262 1 chr1:1645126-1645262 3.25 193 263 1.360 CDK11B 11 11 chr1
3 3 s1 0.9927666 13 665 665 duplication 1 3469433 3469593 1 chr1:3469433-3469593 2.69 145 197 1.360 ARHGEF16 5 5 chr1
4 4 s1 0.9927666 13 1842 1843 deletion 2 12917693 12920413 1 chr1:12917693-12920413 28.60 936 584 0.624 PRAMEF7 1 2 chr1
5 5 s1 0.9927666 13 1846 1848 deletion 3 13049808 13053659 1 chr1:13049808-13053659 33.00 1679 995 0.593 PRAMEF27 1 3 chr1
6 6 s1 0.9927666 13 1851 1852 duplication 3 13175281 13198906 1 chr1:13175281-13198906 66.20 1019 2081 2.040 PRAMEF9 1 2 chr1
6.1 6 s1 0.9927666 13 1853 1853 duplication 3 13175281 13198906 1 chr1:13175281-13198906 66.20 1019 2081 2.040 PRAMEF13 1 1 chr1
7 7 s1 0.9927666 13 1856 1856 deletion 1 13260222 13260803 1 chr1:13260222-13260803 18.60 321 136 0.424 PRAMEF5 1 1 chr1
8 8 s1 0.9927666 13 2208 2208 deletion 1 16575157 16575229 1 chr1:16575157-16575229 4.29 440 300 0.682 NBPF1 4 4 chr1
9 9 s1 0.9927666 13 2217 2217 deletion 1 16587050 16587261 1 chr1:16587050-16587261 15.30 640 350 0.547 NBPF1 13 13 chr1
10 10 s1 0.9927666 13 2219 2221 deletion 3 16588849 16592021 1 chr1:16588849-16592021 8.34 1302 933 0.717 NBPF1 15 17 chr1
11 11 s1 0.9927666 13 2865 2865 duplication 1 21480051 21480223 1 chr1:21480051-21480223 8.55 225 354 1.570 NBPF3 10 10 chr1
12 12 s1 0.9927666 13 2865 2867 deletion 2 21480051 21481769 1 chr1:21480051-21481769 5.79 752 610 0.811 NBPF3 10 12 chr1
13 13 s1 0.9927666 13 3042 3042 deletion 1 22003003 22003088 1 chr1:22003003-22003088 5.81 169 98 0.580 CELA3A 2 2 chr1
14 14 s1 0.9927666 13 5325 5325 duplication 2 39763689 39763791 1 chr1:39763689-39763791 5.27 477 629 1.320 PPIE 2 2 chr1
14.1 14 s1 0.9927666 13 5326 5326 duplication 2 39763689 39763791 1 chr1:39763689-39763791 5.27 477 629 1.320 BMP8B 3 3 chr1
15 15 s1 0.9927666 13 5329 5329 duplication 1 39788152 39788485 1 chr1:39788152-39788485 2.93 442 574 1.300 BMP8B 1 1 chr1
16 16 s1 0.9927666 13 8814 8814 deletion 1 89010904 89011116 1 chr1:89010904-89011116 3.95 257 170 0.661 GBP3 4 4 chr1
17 17 s1 0.9927666 13 9677 9686 duplication 12 103571603 103618100 1 chr1:103571603-103618100 57.20 4654 6320 1.360 AMY2B 1 10 chr1
17.1 17 s1 0.9927666 13 9687 9688 duplication 12 103571603 103618100 1 chr1:103571603-103618100 57.20 4654 6320 1.360 AMY2A 1 2 chr1
I guess I don't need to provide 200 samples or more to get higher resolution, as DECoN takes only small subset of references? (like 13 on below picture)?
