Data are from Swebic I, selected groups by prospective suicide attempts. Matched on litihum treatment, age and sex. The groups are also chose so that the 0 suicide group also have no past attempts.
Define phenotypic metadata
How many suicide attempt samples are there?
So there are 398 suicide attempts all in all. All suicide attempts mentioned here are prospective, and here we look at number of days between sampling and suicide attempt.
What does number codec mean in drugs?
Two samples have all NA's --> exclude them!
Were they QC=Warning?
Don't really know what that QC is about...
There have been some discussion about LOD's in the group. Let's look at distributions for most every marker we got. May always be good to return to a histogram as well
Perhaps I should unit variance scale as well...
Comparing scaled with log2(scaled)
Scale and log2 produces a left tail(?)
To estimate the variance in the panel, and to find any unsupervised clustering
Make sure this PCA model scales data!
Looking surprisingly interesting!
Plot the 17 significant markers
Det är inte mycket signal här!
Only two proteins remain!
A simple supervised clustering method should find some significances here!