Hello, I am analyzing human serum data for biomarker discovery of a large scale study. I did not conduct the experiments, they were done by a lab tech. The instrument used was Thermo Qexactive.

The serum samples were run in 4 batches. Pooled QCs run with every batch. The first two batches were run in September, the third in November and forth in December. Now, across the batches, I can see an overall RT time drift of 50 seconds. I have manually verified the major peaks and they are the same m/z peaks. The batch effect is nonlinear in nature as well.

Why do you think this RT shift has happened?