Question: Best way of presenting "absolute" expression values (edgeR)
0
5.2 years ago by
Sindre Lee100
Sindre Lee100 wrote:
Hi! We want to classify a new type of glands by ranking genes by expression level using RNAseq. We don't have any good controls, so we just want to see a ranked list of genes. I have used Cufflinks RPKM values, but if I want to use edgeR, is this a valid way of doing it using featureCounts: fc <- featureCounts(files=targets$Targets,nthreads=8, isGTFAnnotationFile=TRUE, GTF.attrType="gene_id", GTF.featureType="exon", useMetaFeatures=TRUE, annot.ext="genes.gtf") x <- DGEList(counts=fc$counts, genes=fc$annotation) expr <- calcNormFactors(x) expr_norm <- rpkm(expr, log=FALSE,gene.length=x$genes\$Length) # Getting gene length from FeatureCounts, using rkpm() in the edgeR package, not Rsubread.. Then just write out this table.. Thanks!
rnaseq edger • 1.3k views
modified 5.2 years ago • written 5.2 years ago by Sindre Lee100
Hi! I know a lot of people do the following: 1. Take a list of differentially expressed genes 2. Fetch the FASTA files for protein coding genes 3. Predict if secretory by using SignalP My question is, does it exist a list/database of known/predicted secretory protein coding genes from hg19? That would be much more efficient than many people predicting the same proteins a lot of times..
Answer: Best way of presenting "absolute" expression values (edgeR)
0
5.2 years ago by
Gordon Smyth37k
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
Gordon Smyth37k wrote:

Yes, your code is fine for getting normalized RPKM from featureCounts and edgeR.

Your code is similar to the public case study: http://bioinf.wehi.edu.au/RNAseqCaseStudy

In the latest version of edgeR, you can even simplify the code to

expr_norm <- rpkm(expr)

Gordon