Research

Protein PRS Calibration Dashboard

Reference calibration for DNA-derived protein prediction models used as a supporting pathway layer in Helix reports. The dashboard shows model quality, reference percentile spread, coverage, and panel mapping before report-grade filtering.

Protein models
1,636

All protein PRS models available in the reference set.

R2 >= 0.50
98

Models with more genetic than residual variation by this threshold.

Avg coverage
94.1%

Average model coverage in the demo reference sample.

Extreme tails
138

Models in the <=10th or >=90th percentile tails before category review.

Reference Percentile Distribution

Protein-level genetic predictions after reference percentile calibration.

0-10
95
models
10-20
194
models
20-30
196
models
30-40
234
models
40-50
212
models
50-60
201
models
60-70
214
models
70-80
155
models
80-90
94
models
90-100
41
models

Model R2 Distribution

Higher R2 means the DNA model explains more of the protein-level prediction.

<0.10
368
models
0.10-0.20
565
models
0.20-0.30
381
models
0.30-0.40
149
models
0.40-0.50
75
models
>=0.50
98
models

Coverage Distribution

How much of each published protein model was available for this demo sample.

<80
3
models
80-90
58
models
90-95
1,047
models
95-99
518
models
99-100
10
models

Panel Mapping

Current high-level pathway labels used for report evidence context.

Immune
77
models
Cancer
26
models
Metabolic
36
models
Cardiovascular
42
models
Neurological
15
models
Median percentile
45.2%
Median R2
0.178
R2 >= 0.40
173

Report-Grade Tail Examples

Examples that pass R2 >= 0.50 and sit near either reference tail. Reports use these as pathway context, not as standalone blood-test results.

R2 floor: 0.50
CTSH
CTSH
4.0%
percentile
R2 0.62coverage 95.3%
IL1RAP
IL1RAP
94.2%
percentile
R2 0.71coverage 95.0%
ICAM2
ICAM2
7.6%
percentile
R2 0.55coverage 93.0%
MICB_MICA
MICB_MICA
91.0%
percentile
R2 0.72coverage 95.0%
MAN2B2
MAN2B2
9.6%
percentile
R2 0.60coverage 95.1%
View demo reportRead methodology notes