Wired Fit Curve

[Yuxiiii]

Hello Yafeng,

Thanks for the wonderful tool! I used DEqMS for analysing my label free proteomics dataset with a simple design con1 vs con2. I saw a poor fit curve (curve droped and disappear around PSM count 5). I have used protein counts and used a relatively strict filter to keep the proteins only presented in all samples. and i manually checked that there was no NA in fit3$count.

Could you please help about the possible reason of this curve? And even with this Curve, i have quite decent number for differential expression test. Do you think if these results are trustable?

Thank you very much in advance,
Yuxi

[Yuxiiii]

Hello again,

I found out I used MS2 quantity as count in previous pipeline which might be wrong. After changed to PG.NrOfPrecursorsUsedForQuantification (since i don't have PG.UsedPeptides), the fitting curve seems to be better. May i ask your opinion about this?

Best,
Yuxi

[Yuxiiii]

Here are 2 questions regarding Label free dataset:

1. In the vigenettes for LFQ part, you only used log trasnformed data as input of limma part. Do you recommand to do equalMedianNormalization() on the dataset as well?
2. I noticed you added pseudocount 1 to all proteins in count to get rid of 0. Have you tried to select min value of the non-zero count number instead of adding pseudocount?

Best,
Yuxi