It should relate to the mean of normalsed counts across all samples, not just those that you are comparing via results.
Here is the proof:
Check mean of normalised counts across entire dataset
apply(norm, 1, mean)[1:10]
A1BG_protein_coding A1CF_protein_coding A2ML1_protein_coding
0.22956887 3.04713840 1.60183727
A2M_protein_coding A3GALT2_protein_coding A4GALT_protein_coding
0.04976802 0.17302601 7.88745689
AAAS_protein_coding AACS_protein_coding AADAT_protein_coding
252.53146033 279.67821698 0.42314705
AAED1_protein_coding
45.81686538
Check mean reported in different results objects:
deg8$baseMean[1:10]
[1] 0.22956887 3.04713840 1.60183727 0.04976802 0.17302601
[6] 7.88745689 252.53146033 279.67821698 0.42314705 45.81686538
deg9$baseMean[1:10]
[1] 0.22956887 3.04713840 1.60183727 0.04976802 0.17302601
[6] 7.88745689 252.53146033 279.67821698 0.42314705 45.81686538
deg10$baseMean[1:10]
[1] 0.22956887 3.04713840 1.60183727 0.04976802 0.17302601
[6] 7.88745689 252.53146033 279.67821698 0.42314705 45.81686538
Check that all are equal:
table(apply(norm, 1, mean) == deg8$baseMean)
TRUE
18062
table(apply(norm, 1, mean) == deg9$baseMean)
TRUE
18062
table(apply(norm, 1, mean) == deg10$baseMean)
TRUE
18062
Kevin