Good practice in concatenating data

I have a question regarding good practice in concatenating different runs. I usually have 2-3 runs for the same subject that i’d like to concatenate. For the moment i was concatenating directly the raws before filtering and downsampling so that i could extract annotations from the stim channel having the whole thing. I saw that it automatically marks the edge so i thought that was enough for avoiding having filtering issues, is that correct or is it always better to first preprocess and then merge the datasets when we epoch?
Thank you in advance

Hello @Hal42 and welcome to the forum!

My advice would be to process each raw individually before concatenation. This will also allow you to mark different sets of channels as bad and later — at the time of concatenation — decide how to deal with this, e.g. with the help of equalize_bads()

Best wishes,