Table 5.
Previous models for red hair prediction
| Publication | Year | Phenotype | N | P | MC1R variants | Other variants | AUROC | Other metric |
|---|---|---|---|---|---|---|---|---|
| Grimes et al. (18) | 2001 | Red and auburn | 197 | 0.274 | 14 | NA | NA | 20.960 |
| Branicki et al. (34) | 2007 | Red and blonde-red | 184 | 0.410 | 32 | NA | NA | 20.975 |
| Branicki et al. (35) | 2007 | Red | 390 | 0.240 | 43 | NA | NA | 20.960 |
| Sulem et al. (75) | 2007 | Red | 56918 | 0.055 | 32 | NA | NA | 60.700 |
| Walsh et al. (56) | 2013 | Red, blonde-red and auburn | 1551 | 0.088 | 74 | 811 | NA | 90.800 |
| Branicki et al. (37) | 2011 | Red, blonde-red and auburn | 385 | 0.249 | 102 | 1111 | 0.90 | 12 |
| Walsh et al. (6) | 2014 | Red, blonde-red and auburn | 1601 | 0.085 | 1311 | 144 | 0.92 | NA |
| Sochtig et al. (41) | 2015 | Red, blonde-red and auburn | 605 | 0.14 | 155 | 167 | 0.94 | 17 |
| Caliebe et al. (39) | 2016 | Red tint | 400 | 0.31 | 3 | NA | 0.75 | 18 |
| Siewierska-Gorska et al. (42) | 2017 | Red and blonde-red | 186 | 0.24 | 3 | 19 | 0.84 | 20 |
| Hysi et al. (40) | 2018 | Red | 2115 015 | 22 | 238 | 24268 | 250.87; 260.84 | 270.35 |
N, sample size; P, red hair prevalence in the sample.
1rs312262906, rs555179612, rs1805006, rs11547464
2Precision for variant homozygous or compound heterozygous redheads
3rs1805007, rs1805008
4rs1805007, rs1805008, rs11547464
55704 Icelanders and 1214 Dutch
6Precision at 0.50 classification threshold
7rs201326893, rs312262906, rs1805006, rs11547464
8rs1042602 (TYR), rs4959270 (EXOC2), rs28777 (SLC45A2), rs683 (TYRP1), rs2402130 (SLC24A4), rs12821256 (KITLG), rs2378249 (ASIP), rs12913832 (HERC2), rs1800407 (OCA2), rs16891982 (SLC45A2), rs12203592 (IRF4)
9Multiple linear regression highest probability hair colour category + a model for binary hair colour shade (light/dark) prediction; these two models used to make the final prediction, red-hair prediction accuracy, reported here.
10Combined minor allele count (max. 2) at any of the high-penetrance ‘R’ variants (rs201326893, rs312262906, rs1805006, rs11547464, rs1805007, rs1805008, rs1805009) or low-penetrance ‘r’ variants (rs1805005, rs2228479, rs1110400, rs885479)
11rs12913832 (HERC2), rs12203592 (IRF4), rs1042602 (TYR), rs4959270 (EXOC2), rs28777 (SLC45A2), rs683 (TYRP1), rs1800407 (OCA2), rs2402130 (SLC24A4), rs12821256 (KITLG), rs16891982 (SLC45A2), rs2378249 (ASIP)
12Sensitivity 0.78, specificity 0.95, precision 0.84, negative predictive value 0.93; 0.86 AUC for LASSO model
13rs201326893, rs312262906, rs1805006, rs11547464, rs1805007, rs1805008, rs1805009, rs1805005, rs2228479, rs1110400, rs885479
14rs1042602 (TYR), rs4959270 (EXOC2), rs28777 (SLC45A2), rs683 (TYRP1)
15rs11547464, rs1805006, rs1805007, rs1805008, rs1805009
16rs28777 (SLC45A2), rs35264875 (TPCN2), rs1129038, rs12913832 (HERC2), rs4778138, rs7495174 (OCA2), rs12931267 (FANCA)
17Bayes classification
18Sensitivity 0.19; specificity 0.09; accuracy 0.74; heritability for rs1805007 0.14 and for rs1805008 0.07
19rs16891982 (SLC45A2), rs12913832 (HERC2), rs1800401 (OCA2))
20Sensitivity 0.67; specificity 0.93; accuracy 0.87; positive predictive value (PPV) 0.74; negative predictive value (NPV) 0.90
217291 QIMR (Brisbane Twin Nevus Study, Australian Twin Registry, and Tasmanian Eye Study) and 7724 RS (Rotterdam Study)
22QIMR 0.054, RS 0.031, UKBB 0.047
23rs1805006, rs11547464, rs1805007, rs1805008, rs1805009, rs1805005, rs2228479, rs1110400
25QIMR
26RS
27Heritability