Saturday, January 28, 2023

BIRCH -- Identify those batch effects and fix them!

 


Well....that was a rabbit hole that ended at some old scifi horror show about tree monster. An ad that said the "root of all evil" made me think they probably gave it the correct level of seriousness.

HEY! Remember when we did like 2 condition proteomics? That was cool, because we could run our control, then our treated and spectral count some stuff. No fancy stats, just crap data! (J/k) 

Now, however, we have to do the grown up science with lots and lots of samples. BATCHES of samples, you might say, and now we need to start thinking hard about BATCH EFFECTS.

What's that? Well, that's the differences that aren't your phenotype. Those are the ones that are due to the fact you had a lot more dog dander on yourself when you processed the samples on Monday than the next batch of samples on Friday. Or, if you're in an old dungeon where the HVAC swings your internal temp from 45-95F between those dates....those changes.....

And that brings me back to BIRCH

You can get it here

You need the things above! Your non-normalized file. Your normalized file. A stable internet connection (not sure why) and the software at that Github.

It'll try to sort out what effects were caused by that dog dander when it was 95F in your lab and what was actually due to your CRISPR knockout! 


No comments:

Post a Comment