Table 2 The accuracy of PRS models for predicting ANA positive to SLE progressions in the All of Us biobank

From: Integrating electronic health records and GWAS summary statistics to predict the progression of autoimmune diseases from preclinical stages

Model

\({R}^{2}\)

AUPRC

AUC

CC-Lassosum

0.033 (0.015, 0.052)

0.112 (0.088, 0.15)

0.568 (0.506, 0.628)

CC-LDpred2

0.01 (0, 0.021)

0.096 (0.077, 0.123)

0.540 (0.482, 0.596)

CC-PRS-CS

0.032 (0.014, 0.05)

0.11 (0.087, 0.146)

0.566 (0.508, 0.625)

PROG-Lassosum

0.001 (−0.002, 0.004)

0.093 (0.076, 0.127)

0.519 (0.465, 0.575)

PROG-LDpred2

0.019 (0.005, 0.033)

0.096 (0.079, 0.12)

0.554 (0.498, 0.606)

PROG-PRS-CS

0.01 (0, 0.021)

0.095 (0.078, 0.119)

0.543 (0.488, 0.592)

MTAG-Lassosum

0.001 (−0.002, 0.003)

0.091 (0.074, 0.116)

0.516 (0.455, 0.573)

MTAG-LDpred2

0.005 (−0.002, 0.012)

0.09 (0.075, 0.11)

0.533 (0.479, 0.583)

MTAG-PRS-CS

0.003 (−0.003, 0.008)

0.091 (0.076, 0.113)

0.528 (0.471, 0.576)

GPS-Lassosum

0.044 (0.023, 0.065)

0.124 (0.091, 0.171)

0.568 (0.511, 0.622)

GPS-LDpred2

0.037 (0.018, 0.056)

0.117 (0.089, 0.167)

0.566 (0.513, 0.623)

GPS-PRS-CS

0.042 (0.021, 0.062)

0.119 (0.09, 0.163)

0.566 (0.51, 0.623)

TL-PRS-Lassosum

0.028 (0.011, 0.044)

0.102 (0.084, 0.131)

0.561 (0.507, 0.612)

TL-PRS-LDpred2

0.007 (−0.002, 0.016)

0.093 (0.077, 0.118)

0.527 (0.477, 0.578)

TL-PRS-PRS-CS

0.027 (0.011, 0.044)

0.100 (0.083, 0.126)

0.558 (0.509, 0.605)

STACKING-Lassosum

0.034 (0.016, 0.053)

0.111 (0.088, 0.148)

0.570 (0.51, 0.63)

STACKING-LDpred2

0.028 (0.011, 0.046)

0.103 (0.082, 0.131)

0.558 (0.499, 0.611)

STACKING-PRS-CS

0.039 (0.019, 0.059)

0.112 (0.089, 0.149)

0.569 (0.51, 0.626)

MVL

0.0046 (−0.0025, 0.012)

0.090 (0.075, 0.113)

0.529 (0.476, 0.579)

  1. We report the Nagelkerke’s \({R}^{{{\boldsymbol{2}}}}\), AUPRC (area under the precision recall curve), and AUC (area under the receiver operating characteristic curve). The 95% confidence intervals for the estimates are listed in the parenthesis. The top two methods for each metric are displayed in bold and italic font.