Are unequal pool sizes for RNA-seq acceptable?

eds285 · 10-01-2014, 10:32 PM

Background:
I want to find differentially expressed genes between wild type (WT) and knockout (KO) mice of mixed background using RNA-seq (Illumina HiSeq, 50M reads standard at my core). There will be anywhere between 1-4 KO and 1-4 WT animals per litter.

Because it is difficult to isolate enough cells from one animal, I need to pool cells from WT littermates and KO littermates. I need 2 animals minimum, but 3 and up is preferable. My thinking is that each litter will be one biological replicate, and I plan to sequence RNA from 3-5 litters (3-5 biological replicates). Because the number of KO vs. WT animals within a litter varies, I may have more of one genotype than the other.

Question:
If I had this scenario:
Bio rep 1: Pool of 3 WT and Pool of 2 KO
Bio rep 2: Pool of 2 WT and Pool of 2 KO
Bio rep 3: Pool of 2 WT and Pool of 4 KO

...or any combination thereof, would this be a statistically poor design? Should I keep the number of animals in each pool constant for all biological replicates? Only within biological replicates?

Another consideration: Will this be representative of only one mating cage? Cost/time is limited to maybe one time on the sequencer, two chips.

dpryan · 10-02-2014, 12:25 AM

In the age of single-cell RNAseq, there's little remaining reason to pool multiple samples like that.

BTW, I'm glad you're taking litter into account. A lot of people don't do that and then their results end up being do to a litter effect.

eds285 · 10-03-2014, 11:09 AM

So you're saying what exactly? Pool size doesn't matter? Keep pool size equal at the minimum of 2 and the sequencing can take care of the rest?

dpryan · 10-03-2014, 02:20 PM

I'm saying pooling is a last resort. Try to find someone local doing single-cell sequencing and you'll be able to both increase your N and eliminate pooling in one go.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
insert sizes for RNA-seq	luc	RNA Sequencing	6	08-13-2014 11:07 AM
Chip-seq tags of unequal length - is this a problem for macs?	feralBiologist	Bioinformatics	1	01-16-2014 09:24 AM
Group sizes in RNA-seq DE	moriah	Bioinformatics	2	08-13-2012 02:23 PM
How many reads are acceptable from an RNA seq experiment	pettervikman	Bioinformatics	23	02-09-2012 05:45 AM

10-01-2014, 10:32 PM	#1
eds285 Junior Member Location: NYC Join Date: Oct 2014 Posts: 2	Are unequal pool sizes for RNA-seq acceptable? Background: I want to find differentially expressed genes between wild type (WT) and knockout (KO) mice of mixed background using RNA-seq (Illumina HiSeq, 50M reads standard at my core). There will be anywhere between 1-4 KO and 1-4 WT animals per litter. Because it is difficult to isolate enough cells from one animal, I need to pool cells from WT littermates and KO littermates. I need 2 animals minimum, but 3 and up is preferable. My thinking is that each litter will be one biological replicate, and I plan to sequence RNA from 3-5 litters (3-5 biological replicates). Because the number of KO vs. WT animals within a litter varies, I may have more of one genotype than the other. Question: If I had this scenario: Bio rep 1: Pool of 3 WT and Pool of 2 KO Bio rep 2: Pool of 2 WT and Pool of 2 KO Bio rep 3: Pool of 2 WT and Pool of 4 KO ...or any combination thereof, would this be a statistically poor design? Should I keep the number of animals in each pool constant for all biological replicates? Only within biological replicates? Another consideration: Will this be representative of only one mating cage? Cost/time is limited to maybe one time on the sequencer, two chips.

10-02-2014, 12:25 AM	#2
dpryan Devon Ryan Location: Freiburg, Germany Join Date: Jul 2011 Posts: 3,480	In the age of single-cell RNAseq, there's little remaining reason to pool multiple samples like that. BTW, I'm glad you're taking litter into account. A lot of people don't do that and then their results end up being do to a litter effect.

10-03-2014, 11:09 AM	#3
eds285 Junior Member Location: NYC Join Date: Oct 2014 Posts: 2	So you're saying what exactly? Pool size doesn't matter? Keep pool size equal at the minimum of 2 and the sequencing can take care of the rest?

10-03-2014, 02:20 PM	#4
dpryan Devon Ryan Location: Freiburg, Germany Join Date: Jul 2011 Posts: 3,480	I'm saying pooling is a last resort. Try to find someone local doing single-cell sequencing and you'll be able to both increase your N and eliminate pooling in one go.