edgeR User Guide batch effect question

Question

[Assa Yeroslaviz]

Hi, In the edgeR use guide in section 4.2.5 it is mention that for further examining the consistency of the three replica the predictive logFC are computed.

than the correlation of the Time1-3 with the three different replica is shown. these are all positively.

I was wondering what this exactly means.

I have done the same analysis with my own data and got also negative values:

                       Replica1:TreatVeh Replica2:TreatVeh Replica3:TreatVeh
Replica1:TreatVeh        1.00000000         0.2891324        0.04387087
Replica2:TreatVeh        0.28913237         1.0000000       -0.13277416
Replica3:TreatVeh        0.04387087        -0.1327742        1.00000000

What does these values mean for the relationship between samples 2 and 3?

also, my MDS plot looks a bit different (not as clean, but surely there is some kind of batch effect in the data):

I would appreciate any advice as to how to continue with the analysis

thanks Assa