Hi,
I have RnaSeq expression in RPKM, and I would like to run an eQTL analysis in R. However just running a linear regression between the RPKM values as they are and genotypes does not seem to give me the correct results (everything comes out as significant).
I must be missing some steps for QC and prep of the data before linear regression. For example maybe I should be using log of RPKM, or maybe raw, or log10, or log2. Also should the data be quantile normalized?
What is the starndard process before doing a linear association for eQTL analysis using RnaSeq RPKM values?
I can't find good guidance on how to prepare data (i.e. sample QC and gene QC required) and run an eQTL analysis using RnaSeq RPKM values in R. If there are any papers or any manuals in R that can help understand the steps to do this analysis, or if anybody knows how I can learn the steps it would be of great help.
Thank you
0 answers
No answers yet.
Log in to answer this question.
-
Seeking Guidance on Standardization Methods for eQTL Analyses with Diverse Preprocessing Techniques
written by DS •
-
Batch correction for eQTL analysis
written by maximal_life •
-
software and publicly available datasets for eQTLs?
written by User 7754
-
RNA seq analysis using linear regression
written by evelyn
-
RNAseq Normalization for eQTL Mapping?
written by QVINTVS_FABIVS_MAXIMVS
-
GTEx RNA sequencing data in RPKM
written by kelly.wang135
-
eQTL analysis on TCGA RNAseq data and SNP6 data
written by vakul.mohanty
-
RNA Seq Time Regression
written by tucanj
-
Expression Quantitative Trait Loci Analysis Using Rpkm In R
written by User 7754
-
Feature Selection Methods For Rnaseq Data?
written by antass •
(1) Do you really have to use RPKM values? The normal tools would work better with counts. (2) How many samples are we talking about here? (3) Can you give us an example of the regression command you're running (presumably
lm(...))? (4) Whether quantile normalization will make sense depends a bit on the experiment.Thank you for your reply. I only have RPKM so I have to use these. I am also not sure how much QC was done on these samples, so would still have to go through that. I have 100 samples, and the regression command is using SnpStats snp.rhs.estimates with RPKM raw expression for now as outcome and the SNP genotypes as explanatory variable: RPKM ~ genotype. I am not sure what other info is needed since I don't really understand what the quantile normalization would depend on...