PartiotionedPRS custom scripts

In this repo:

1. Penalized regression models (+ correlation plots)

2. Gene-mapping of identified loci

3. Calculating, goodness-of-fit and association analysis of overall and partitioned PRS (pPRS)

4. bNMF soft-clustering algorithm results for generating pPRS

Note

You need an existing access to UK Biobank genotypic and phenotypic data. Preferentially to be run in a Windows env with a WSL (Windows Subsystem for Lunux) installed. All dependent Python or R packages should be installed separately

Tip

Each phenotype should be a struct with necessary fields: eid, rawUKB, numericFlag, tag. Fields info, source, and termMeaning can be left empty. exeid field includes exclusion criteria (if any). eid for a binary trait contains only ID of cases with that particular trait.

UKB_STRUCT_ALL

struct with fields:

           info: [1×1 struct]
            eid: [13884×1 double]
         rawUKB: [13884×1 string]
         source: [13884×1 string]
    numericFlag: 0
          exeid: [906×1 double]
             ex: [1×1 struct]
    termMeaning: [13884×1 string]
            tag: "Btrait"

% for a continuous trait:

UKB_STRUCT_ALL = 

  struct with fields:

           info: [1×1 struct]
            eid: [274350×1 double]
         rawUKB: [274350×1 double]
    numericFlag: 1
            tag: "cTrait"
    termMeaning: ''