How to design qPCR primers to validate RNA-seq experiment?

Question

[brett.vanderwerff]

Hello,

I did a bulk mRNA-seq experiment where I treated human cells in vitro with control of drug. I am only interested in determining differential gene expression at the gene level between these groups. I isolated RNA from treated cells and used poly-A selection and did sequencing on the illumina platform. I aligned the raw reads to the human genome using STAR. I used the featureCounts subpackage of subread to to get a count of reads that overlap with genes, the resulting gene count matrix is what I fed into edgeR to get differential gene expression results.

Now I would like to design some primers against a few select genes for qPCR to validate my differential gene expression results.

I am familiar with using Primer-BLAST, but I have previously only used it to design primers that span the exon-exon junction of a given transcript isoform by supplying the NCBI Reference Sequence of that isoform to the Primer-BLAST tool.

If I follow this approach I am afraid of designing primers that span the exon-exon junction of an isoform I don't even have in my sample.

I have also produced the equivalent of an "exon-count-matrix" with featureCounts by counting at the exon level rather than the gene level. I guess I could design primers to bind within a specific exon that I know is expressed, but then I worry about amplifying contaminating genomic DNA (which I realize might not even be an issue because I did a genomic DNA elimination step in my RNA isolation).

How do you usually go about designing primers for qPCR based RNA-seq validation?

Thank you.

[Kevin Blighe]

A previous answer here: A: primer design to discriminate sequences in intron-exon MSA

[Malcolm.Cook]

What aspect of the mRNA-seq results are you seeking to validate? If you are trying to design primers to recapitulate the results of an analysis of the mRNA-seq, you should describe the analysis (and the experiment/samples/sequencing).

That said, primer3 has lots of design options when you get to actual primer design part.

[brett.vanderwerff]

Yes I am trying to validate analysis results, I was not sure how much detail was needed, In apologize for that I have updated my questions to better describe my pipeline.

Answer 1

[biplab]

You can use primer3 for designing primer. http://bioinfo.ut.ee/primer3-0.4.0/

[brett.vanderwerff]

Thank you for responding. I have used Primer3, but my question is less about what primer tool to use and more about what transcript for a particular gene I should be designing my primers against.

Answer 2

[Malcolm.Cook]

Since your analysis of RNASeq was "at the gene level" (trying to be blind to alternative splicing), you might seek for your qPCR to similarly be so. Since you enriched for mature mRNA (poly-A selection), you might design your primers to target amplifying a constitutively expressed spliced product.

By constitutive expressed I mean "identically present in every isoform" (adopting definition from Statistics and Data Analysis for Microarrays Using R and Bioconductor)

So, your primers must fall in the constitutive flanking exonic segments of a constitutive intron.

By such a design, your expression readout should not be altered by a drug effect which alters the proportions isoforms without changing overall (primary mRNA) expression level.

Also, your concern that "I am afraid of designing primers that span the exon-exon junction of an isoform I don't even have in my sample." should be mitigated by this approach since, by design, every isoform will take the junction and so the PCR product of any isoform should be identical.

Of course, you will have other design considerations, such as avoiding regions of, e.g. known SNPs or homology to other coding genes.

Note: not every spliced gene has a "constitutively expressed spliced product". By my analysis, in human hg19 about 20% of spliced genes lack at least one constitutive intron.

If you're doing this only for "a few select genes", then you can eyeball GFF tracks in a genome browser to find whether a candidate gene has a constitutive intron and flanking exonic segments to target. Then just take that sequence to your favorite primer design environment.

If you're doing this at scale, the R/Bioconductor GenomicFeatures package provides the exact rudiments you need to perform such an analysis. You can retrieve intronsByTranscript, determine which are constitutive (present in every transcript/isoform) when grouped by gene, and take those that are, along with their flanking disjointExons, as candidate targets. The code to determine which introns are constitutive is a little tricky but after that find the flanking "disjointExons" is pretty straight ahead.