Ok /sci/ my PI just slapped some RNA-seq data on me and I have no idea what to do with it. How do people typically go about getting meaning from the differential expression lists?
>>8786408
What are you majoring in?
>>8786458
Genetics
>>8786408
Bioinformatics algorithms by compeau and pevzner or however their names are spelled is a good intro to computational genomics.
>>8786408
What kind of actual quantitative data do you get from a RNA-seq analysis?
>>8786408
>RNA-seq data on me and I have no idea what to do with it
Neither does anyone else here. Maybe you should state what the problem is.
>>8786591
hmmm thanks I'll look into it
>>8786652
as far as I understand you get read counts of mrna which then get mapped to a genome and then you can get differences in gene expression between a control group and experimental
>>8786657
my problem is after mapping and doing the differential expression I have no idea where to start with so much data
>>8786652
trim at like 30 quality on both ends and run TopHat on it, more reads on a gene = highly expressed gene
>>8786673
>my problem is after mapping and doing the differential expression I have no idea where to start with so much data
you are done, at this point you have the differentially expressed genes so you need to know wtf you are looking for. You can use stuff like gene anthology if you are just looking for an overview of what pathways are most affected
>>8786683
I suspected there was not much I could do without a hypothesis.
>>8786691
Do you have expression patterns over time, or just a control and experimental group?
>>8786700
it's over time
>>8786706
I forgot what the analysis is called, but I believe there is a way to cluster each set of genes into different "expression pattern" assignments
e.g. gene 1 and gene 65 linearly increase over time so they get put into expression pattern 1. Gene 4 and gene 9 exponentially decrease over time, so they get put into expression pattern 2.
Not sure how much this would help, but it's something, right?
>>8786719
Oh that sounds very interesting, thanks
I'll search around for it
>>8786722
Look up some data science/statistical inference techniques. Those areas excel at handling huge amounts of data. They might give you some inspiration.
With mapped reads and diff. abundance, you can do a few things. First, you could try assembling the mapped reads into a transcriptome using your reference genome. Next, you can do functional profiling of your mapped reads by finding the reference genome's associated KEGG accessions. You could then test, for instance, whether expression of some biochemical pathway was different between subjects. You could also map out the genome/chromosome of interest using a circular plot and then overlay expression data on it.
>>8786719
An ordination method like NMDS or PCA could be used to see if the overall expression profiles are different between time points.