Table 1 Prediction accuracy summary (mean ± standard deviation).

From: Application of deep learning algorithm on whole genome sequencing data uncovers structural variants associated with multiple mental disorders in African American patients

Variation types

Prediction accuracy (single diagnosis mental disorder vs controls)

Prediction accuracy (mental disorder vs controls)

Prediction accuracy (hamming loss among 8 disorders)

Prediction accuracy (exactly matches among 8 disorders)

Nonsynonymous SNVs

71.7 ± 1.34%

64.5 ± 1.2%

0.28 ± 0.011

8.8 ± 1.1%

Frameshift SNVs

70.8 ± 1.61%

64 ± 1.4%

0.30 ± 0.007

8.4 ± 0.91%

Stop codon SNV

71.49 ± 1.69%

65.1 ± 0.97%

0.28 ± 0.004

7.2 ± 0.69%

SNVs in UTR region

72.4 ± 1.44%

65.5 ± 1.1%

0.31 ± 0.009

7.6 ± 1.2%

SNVs in ncRNA

72.6 ± 1.52%

65.7 ± 1.3%

0.29 ± 0.009

8.1 ± 1.4%

SNVs in intronic regions

72.8 ± 1.29%

65.7 ± 1.1%

0.28 ± 0.008

8.1 ± 0.98%

SNVs in intergenic regions

73.1 ± 1.23%

64.6 ± 1.1%

0.30 ± 0.006

9.3 ± 0.90%