search

dms-vep / SARS-CoV-2_XBB.1.5_spike_DMS

Other
5 stars 1 forks source link

Averaging serum elections #8

Closed Bernadetadad closed 1 year ago

Bernadetadad commented 1 year ago

@jbloom I'm not sure if something has been changed for the averaging of serum selections, but I can't reproduce the correlations I had before converting to vep-dms-3 pipeline.

Here's an example for sera 493C. Attached is the script that does the averaging between libraries using previous pipeline and I had great correlation (r=0.691) but with the new pipeline here correlation is only r=0.27. I can't find what has exactly has changed. I also attach individual library fits from previous pipeline, which mostly look the same just the Hill curve has slightly changed (e.g. from 3.51 to 3.45 for lib1 and from 3.58 to 3.54 for Lib2).

Same issue with poor averaging correlations applies to all other sera as well. 493C.zip

Bernadetadad commented 1 year ago

It does look like I did previously used spacial regularization, let me then add that to this pipeline to see if it solves the issue

jbloom commented 1 year ago

OK, set it up to use the same parameters as before and then let me know if it still looks different.