qulity control of RNA-seq results

Question

[elmira b]

Hi,

Should be a correlation between number of good reads and RNA concentrations in the result of RNA sequencing?

We used Tophat 2 alignment and HTSeq count to determine number of good reads at each sample.

Any help would be greatly appreciated.

[amirmhzadeh]

What are the criteria for calling a read "good"?

Answer 1

[Devon Ryan]

I assume "good" means "uniquely countable" in this context, since you mention HTSeq.

In general, no, there's not an obvious relationship between concentration and sequencing results. You obviously need enough RNA for the library prep., but the actual quality of the RNA itself (rather, the quality of the transcripts, meaning the degree of degradation and background contamination) is vastly more important.

Answer 2

[Michele Busby]

You'll get more duplicates with lower input because there won't be as many molecules going into the initial reaction. You can go pretty low without it tanking, though, if you're careful about the choice of protocol. Some single cell data looks pretty complex. Also, you can screw up the sequencing run if you don't put enough material on the plate.

We did a paper that looked at different methods of making low input libraries

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3821180/

Some of the metrics will show you the things that can go wrong.

There's lots of ways to screw up a library though. Look at your base qualities. Are they worse in some of the libraries?