Hello,

I have a RNASeq data from organoids derived from 4 patients. For one of the patients, the dataset includes organoids in two conditions, with and without treatment. Technical replicates were included for all organoids but these samples were collected at different timepoints (suffix P + number refer to this timepoints) to reduce variability due to routine organoid care when changing culture media. In addition, I have 3 different batches. Here is the design.

Sample_name Treatment Patient Replicate Batch

mCTO50BP2 No 5 1 run1

mCTO50BP5 No 5 2 run1

mCTO50BP9 No 5 3 run1

mCTO50P2 Yes 5 1 run1

mCTO50P3 Yes 5 2 run1

mCTO50P6 Yes 5 3 run1

mCTO66S3-P3 No 6 1 run2

mCTO66S3-P7 No 6 2 run2

mCTO66S3-P9 No 6 3 run2

RTO2P5 No 8 1 run1

RTO2P8 No 8 2 run1

RTO7P2 No 9 1 run1

RTO7P7 No 9 2 run1

RTO7-P10 No 9 3 run3

My questions: 1) To perform multiple comparisons

1a .- mCTO50 vs mCTO50B

1b.- (RTO2 + RTO7 + mCTO50) vs (mCTO50B + mCTO66S3)

1c.- RTO7 vs (RTO2+mCTO66S3+mCTO50+mCTO50B)

Should I load into DESeq2 all the samples in the same matrix, construct one model and then perform different contrasts or should I load just the samples that I want to analyze together one at a time?

2) To compare the organoids of patient 5 between them (Treatment Yes Vs Treatment No), and explore the differences in transcription for this patient (what some authors name "personalome"), would it be fine if I treat each of these samples as independent samples, i.e., not collapsing the replicates?

Thanks very much in advance.

Best regards,

Sheila

In the FAQ we discuss (1).

Re (2) we do not recommend to collapse biological replicates. Only technical replicates, which are when the same library is sequenced deeper.