/NOMOGRAMS

Building Genetics Informed Nomograms from MRI Brain Scans

Primary LanguageR

NOMOGRAMS

Project Goal / Idea : To build better Nomgrams by taking genetic information into account.

Steps in the script:

1- Read in the UKBB table of 500k subjects 2- Exclude those without imaging 3- Exclude those who have a history of head trauma or substance abuse

Now looking at Hippocampal volume: 4- Exclude outliers (more than 5 MAE's away from mean). (Do this twice. Once for each hemisphere) 5- Regress out the effect of: Head Scaling and Date of MRI Scan (Do this for each hemisphere seperately)

Using these corrected volumes: 6- Generate the nomogram. (Sliding quantile window analysis to produce percentile curves)

For PRS generation/analysis: 1- Create a .pheno file with 3 colmns: IID and FID are the "eid"s from the ukb table and a pheno column which contains the sum of the left and right HV (corrected)

2- Using that file, and the UKBB plink files, and the ENEGMA GWAS files, input them into PRSsice to find some Polygenic Risk Scores (PRS)

3- Back in R, read in the files and find the association between HV and the PRS scores found at Different thresholds with a linear regression model. Bar plot all results.