Calculate Polygenic Risk Scores (PRS)

[Nicklas Nordborg]

With ticket #1575 we get a lot of genotypes for the Oncoarray data. We would like to be able to calculate Polygenic Risk Scores from this data set. To begin with the PRS with 313 variants that is described here ​https://pubmed.ncbi.nlm.nih.gov/30554720/

The implementation idea is that it should be relatively easy to filter the imputed VCF so that only the data for the 313 variants remain. We could save this as a separate file and include the PRS weights in it. Then it would be relatively easy to calculate the final score and store it as an annotation. But we need to investigate how to handle variants with a lower INFO (quality) score.

[Nicklas Nordborg]

In 7609:

References #1576: Calculate Polygenic Risc Scores

Added an item list for Oncoarray items and a link on the front page.

[Nicklas Nordborg]

In 7621:

References #1576: Calculate Polygenic Risk Scores (PRS)

Started to implement the "Start Polygenic Risk Score calculation". It should be possible to select genotype call items, but not to submit jobs.

[Nicklas Nordborg]

In 7622:

References #1576: Calculate Polygenic Risk Scores (PRS)

Added <polygenic-risk-score> section to reggie-config.xml for defining available risk score definitions. Currently we only have the PRS313.

[Nicklas Nordborg]

In 7623:

References #1576: Calculate Polygenic Risk Scores (PRS)

Added a software and job subtype for PRS.

[Nicklas Nordborg]

In 7626:

References #1576: Calculate Polygenic Risk Scores (PRS)

Started to implement the PRS calculator. Jobs can be submitted to the cluster but the script will only copy the source VCF file so far.

[Nicklas Nordborg]

In 7627:

References #1576: Calculate Polygenic Risk Scores (PRS)

The script will now extract relevant variants and save them to a separate VCF file. The file is normalized, filtered and annotated with the PRS weights so that it should be possible to calculate the score with the information that is in the VCF.

The score is not yet calculated.

[Nicklas Nordborg]

In 7628:

References #1576: Calculate Polygenic Risk Scores (PRS)

Added a python script that calculates the PRS. The script is really strict and require that all information exists or it will end with an error. We may have to relax things in the future if we find cases that fails.

[Nicklas Nordborg]

In 7639:

References #1576: Calculate Polygenic Risk Scores (PRS)

The resulting VCF file and scores are imported into BASE.

[Nicklas Nordborg]

In 7640:

References #1576: Calculate Polygenic Risk Scores (PRS)

Implemented auto-confirmation which in this case only means that a failed job is re-scheduled for another try.