In the table olink_prot_exp column olink_prot_exp I find the values “Pass” and “Warning”. For those samples labeled “Warning” does that mean some can be included but with caution? Or should we filter all of them out?
And a second semi-related question, in the “Systems view…” paper there is mention of excluded some proteins from the analysis due to values below the level of detection (23), duplicates between the olink panels (12) and also for proteins with a coefficient of variation over 10% eventually reduce the number of proteins measured to 209. Will part or all of this preprocessing also be included in the database? As a quick check I filtered the olink table by the “Warning” flag using this code:
Samples that do not pass the QC are indicated in “npx_quality_control” column as “Warning”. Data from these samples should be treated with caution [1].
We included all data from Olink experiment into Olink_prot_exp table without any pre-processing/filtering. Now, users can decide their own pre-processing/filtration scheme for downstream analysis. One of such example is as you mentioned analysis performed in “Systems view…” paper where standard Olink guidelines are followed [2]. And hence, you can see “olink_prot_exp” table contains two specific columns: “npx_limit_of_detection”, and “npx_quality_control” in order to decide on measures for pre-processing.
Regarding your last question, its not necessary that a protein would not pass QC against all samples. So, there can be set of proteins passing QC for some samples and not passing for other samples and vice versa.
I hope this helps but please do get in touch if you have any further
questions.
