Batch effect from to sequencing data at differents depths
2
1
Entering edit mode
ary.lech ▴ 10
@1b78463a
Last seen 6 months ago
Mexico

Hello,

I have the following RNA-seq experimental design:

  • Control: 3 biological replicates from ther first sequencing + resequencing from the 3 same samples
  • Treatment1: 3 biological replicates from ther first sequencing + resequencing from the 3 same samples
  • Treatment2: 3 biological replicates from ther first sequencing + resequencing from the 3 same samples
  • The resequencing was made from the same libraries but with more sequencing depth (much more depth)

So, technically the 3 resequencing data are technical replicates with a batch effect and the depth effect.

To solve the batch effect i used the function collapseReplicates and then I used the function DESeq from normalization data; which If I'm correct, consider library size. So, the effect of different sequencig depth is considered? or I have to do another kind of normalization?

I have to take another consideration to do the proper analysis? I think a need another consideration but I don't know what

Thank you ,

Sequencing sequencingDeepth BatchEffect DESeq2 • 501 views
ADD COMMENT
1
Entering edit mode

If you have additional metadata on your samples (e.g., sequencing run information, processing dates, etc.), it may be useful to perform a PCA or hierarchical clustering analysis to see how your samples group. For this though, you'll want to make sure to use either the rlog or vst transformations in the DESeq2 package on your count data. If your samples are clustering by sequencing run, processing date, or some other technical variable, you may be able to use COMBAT, RUV-seq, or SVA to account for unwanted variation depending on how the processing was performed. I think at least one of the methods I mentioned allows you to utilize technical replicates to account for batch effects.

I haven't personally used the batch correction methods, but hopefully this gives you a starting point.

ADD REPLY
0
Entering edit mode

Yes, I have additional metadata that includes the sequencing run information. So, I´m going to try with a PCA

Thanks

ADD REPLY
2
Entering edit mode
@james-w-macdonald-5106
Last seen 5 hours ago
United States

There is no batch effect if you use collapseReplicates. It just sums up all the counts for each gene across the resequencing batches (so you end up with three replicates for each group). You then just process as usual, which it appears you have done.

ADD COMMENT
0
Entering edit mode

Yes, I just what i did. It seems that inadvertently I made the right strategy

Thanks

ADD REPLY
2
Entering edit mode
swbarnes2 ★ 1.1k
@swbarnes2-14086
Last seen 2 hours ago
San Diego

So, technically the 3 resequencing data are technical replicates with a batch effect and the depth effect.

No you don't. Running the same library on multiple days does not introduce a batch effect. And there is no "depth effect". Just join all your reads together, just like you would if you had the same sample run in a few different lanes.

ADD COMMENT
0
Entering edit mode

Great!!! i was confused about it but now it´s clearly for me. thanks,

ADD REPLY

Login before adding your answer.

Traffic: 323 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6