Hello,
I need guidance about a time series experiment in which two treatment groups share a common time 0 control. The purpose is to evaluate the effects of hypoxia on differentiation of a specific cell line, based on changes in gene expression over the time points sampled. Cells were grown under normoxia (control condition) and then induced to differentiate under normoxia or hypoxia. Cells were sampled at 12, 24, 36, and 48 hrs after the induction of differentiation. Cells were also collected at time 0 -- just before the switch (or not) to hypoxia. This group serves as a time zero reference for both conditions, but it creates an unbalanced design since time 0 (by definition) only exists for the normoxic condition. I can use only the 4 time points (8 groups) to run the full and reduced model to look for interactions (how hypoxia influences the expected time-dependent changes in gene expression), and to compare normoxia to hypoxia at each time point using contrasts. However, this would preclude our ability to detect the earliest responses to hypoxia (T0 vs. T12). Is there a way to use T0 as a time zero for both treatment groups in DESeq2?
My sample matrix is pasted below.
Sample Treatment Time
S1 Normoxia T0
S2 Normoxia T0
S3 Normoxia T0
S4 Normoxia T0
S5 Normoxia T0
S6 Normoxia T0
S7 Hypoxia T12
S8 Hypoxia T12
S9 Hypoxia T12
S10 Hypoxia T12
S11 Hypoxia T12
S12 Hypoxia T12
S13 Normoxia T12
S14 Normoxia T12
S15 Normoxia T12
S16 Normoxia T12
S17 Normoxia T12
S18 Normoxia T12
S19 Hypoxia T24
S20 Hypoxia T24
S21 Hypoxia T24
S22 Hypoxia T24
S23 Hypoxia T24
S24 Hypoxia T24
S25 Normoxia T24
S26 Normoxia T24
S27 Normoxia T24
S28 Normoxia T24
S29 Normoxia T24
S30 Normoxia T24
S31 Hypoxia T36
S32 Hypoxia T36
S33 Hypoxia T36
S34 Hypoxia T36
S35 Hypoxia T36
S36 Hypoxia T36
S37 Normoxia T36
S38 Normoxia T36
S39 Normoxia T36
S40 Normoxia T36
S41 Normoxia T36
S42 Normoxia T36
S43 Hypoxia T48
S44 Hypoxia T48
S45 Hypoxia T48
S46 Hypoxia T48
S47 Hypoxia T48
S48 Hypoxia T48
S49 Normoxia T48
S50 Normoxia T48
S51 Normoxia T48
S52 Normoxia T48
S53 Normoxia T48
S54 Normoxia T48
This seems to be the same question as asked previously although without specific reference to DESeq2: Time series with two treatments including a common time zero
Yes, it is, but after having used DESeq2, as it's normalization approach is more suitable for my dataset. I am no longer asking if I can load normalized data into the tool; rather, am asking how to handle a time series data set when the time 0 is the reference for both conditions.