Does anybody have a clue how to analyse targeted RNA NGS data with respect to normalization? How to correct for differences in capturing efficiency per transcript? We isolate patient RNA from PAX gene, perform a Sureselect sample prep (oncopanel) and want to see which genes in a patient are differentially expressed. For whole transcriptome analysis we use the "DROP pipeline".
After Quality filtering and adapter trimming, you can use Hisat and stringtie tools for reference genome alignment and Assembly of your transcripts, once you have assembly annotations of your samples in gtf format, you can use this to make DEseq ready files by stringtie tool, followed by use of a python script "prepDE.py", which will used strigtie output and generate a count table for all your transcripts, then you can use this count file further perform DEseq diffrential analysis.
Plz note for targeted you have fetch only your target transcripts from reference genome annotation gtf files and this file you will use as input for assembly.
I do not recommend following Anita Tripathi advice. It is completely irrelevant to what you are asking (and outdated).
I do not recommend using targeted RNA-Seq of a single transcript for expression estimates. If you must, here are my 2 cents.
Normalization of targeted RNA is complicated if not impossible. Two main reasons: 1) differences in capturing efficiency per transcript and per experiment - unknown unless you do extensive testing every time, 2) variable sequencing depth with no reference. The only solution that comes to my mind is adding spike-in molecules with known expression/concentration and using their levels for the normalization. This would, however, most likely require you to design synthetic spike-in molecules with the same target regions as your transcripts (to estimate capture efficiency) in addition to "classic" spike-in molecules (to have a reference for sequencing depth normalization).
I agree with you Jan Oppelt ,
Differential expression from targeted bulk sequencing on a single gene is extremely inaccurate. In future experiments, it is appropriate to use ERCC spike ins or a panel of house keeping genes to normalize for differences in sequencing depth. Here you would need choose house keeping genes that with evenly expressed across all samples, this is still hard to do and know without first performing bulk transcriptome sequencing.
Otherwise, it is probably much cheaper and faster to quantify changes in expression level using qRT-PCR of a single target gene. Changes in gene expression are normalized to the expression two or three house keeping genes. We use GAPDH, ACTB to normalize. You also need to normalize the RNA concentration used as template for RT/qRTPCR reactions.
Finally, de novo transcript assembly is rarely accurate and often leads to the incorrect assembly and quantification of false transcripts.
- Badges
- Science topic
- Similar topics
- Counseling Psychology
- Confusion