How To Deal With Technical And Biological Replicates With Tophat2 For Rna-Seq Data

Question

[Jordan]

Hi,

Let's I have 6 samples - A, B, C and A', B', C'. Here A', B', C' are technical replicates of A, B and C respectively.

And A,B,C are biological replicates. So are A',B',C'. How do I map these samples using tophat2?

According to the website, it's best to map them separately if we want to do differential expression. And I plan to cuffdiff another set of samples (D,E,F) which are similar to A,B,C.

I would normally do something like this for biological replicates,

tophat2 -G annotation.gtf -o sampleA_results Bowtie2/genome  sampleA.fq
tophat2 -G annotation.gtf -o sampleB_results Bowtie2/genome  sampleB.fq
tophat2 -G annotation.gtf -o sampleC_results Bowtie2/genome  sampleC.fq

But with technical replicates, I'm not sure how to proceed. Should I combine the fastq files or should I just do the same command like above on A',B' and C' too? In the end, I plan doing cuffdiff. With bio replicates I can just do the following:

cuffdiff /refs/mm9/Annotation/Genes/genes.gtf -L G1,G2 A.bam,B.bam,C.bam D.bam,E.bam,F.bam

I'm not really sure how to proceed with technical replicates now in the picture. Any help on this would be appreciated.

I'm working on mouse data, single end reads.

Thanks!

Answer 1

[Devon Ryan]

In general, you merge the technical replicates together and map the biological replicates separately. Whether to merge the technical replicates before or after mapping is largely up to you (unless the libraries were constructed independently, in which case the insert size might vary a bit) and will typically make little if any difference.

[Jordan]

So, it's ok if I map them separately and merge them using merge bam from samtools right? And I think they are made from the same library.

[Devon Ryan]

Yeah, that'd be fine.