degraded clinical RNA samples - to seq or not

[chipgal]

Hi,

We have received some human clinical RNA samples. We intend to seq them (Illumina HiSeq) to look for changes in gene expression (protein-coding genes as well as non-coding RNA). We checked the integrity of the RNA and found that quite a few of them are degraded (samples appear as smear on gel, cant even see the 2 ribosomal RNA bands).

The Q is: should we still seq these degraded RNA? I read about the RNase H and Ribo-Zero method which supposedly will work well for degraded RNA samples, but I am not sure if these methods are really good. I could potentially wait for a few more months to get more clinical samples, but of course it will delay the project.

Any advice? Would also like to hear from people who have tired RNase H or Ribo-zero method how their results are like compared to poly-A enriched method.

Thanks!

[blancha]

This is the opinion of a bioinformatics analyst.
No !!!!

It's really hard to make heads or tail out of RNA-Seq from degraded samples.
After having wasted a considerable amount of time trying to analyze degraded RNA-Seq samples, I refuse to do it anymore.
I've asked our sequencing center to advise users not to sequence samples with low RNA Integrity Numbers (RIN).
I've also asked them to provide me with the RIN numbers so I don't waste time trying to understand samples with low RIN.
Samples with degraded RNA will have alignment patterns that appear somewhat random with low correlation between replicates.

As far as ribosomal depletion with the Ribo-Zero kit, I don't share others' enthusiasm for ribosomal depletion versus poly-A enrichment. Yes, with ribosomal depletion, one can get all the non-codings RNAs without poly-A tails, and it may be more appropriate for degraded RNA. However, one also ends up with a lot of "junk", even when the quality of the RNA is good. I find the data from poly-A enrichment "cleaner", and I would recommend it over ribosomal depletion if one is not specifically interested in non-coding RNAs without poly-A tails.

I am familiar with the dilemma with human clinical samples. If the samples are unique and rare, the temptation is great to sequence them even if the quality is low. However, in my experience, trying to make sense of such data is a frustrating exercise.

[Brian Bushnell]

A second opinion. RNA-seq is very difficult and noisy under the best conditions, to the point that it's hard to make a strong or even repeatable conclusion.

Unless you have no choice but to work with low-quality data, your time - and that of everyone who reads your results - is better spent if you obtain high-quality data, rather than making unreproducible claims from data you know to be low-quality.

[kopi-o]

At least review these papers to understand how low RIN might affect quantification and differential expression before you give up hope:

"Sequencing Degraded RNA Addressed by 3' Tag Counting" (http://www.plosone.org/article/info%...l.pone.0091851).

"RNA-seq: Impact of RNA degradation on transcript quantification" (http://www.biomedcentral.com/1741-7007/12/42/abstract#) and

[seqtechno1]

This is the opinion of a bioinformatics analyst.
No !!!!

It's really hard to make heads or tail out of RNA-Seq from degraded samples.
After having wasted a considerable amount of time trying to analyze degraded RNA-Seq samples, I refuse to do it anymore.
I've asked our sequencing center to advise users not to sequence samples with low RNA Integrity Numbers (RIN).
I've also asked them to provide me with the RIN numbers so I don't waste time trying to understand samples with low RIN.
Samples with degraded RNA will have alignment patterns that appear somewhat random with low correlation between replicates.

As far as ribosomal depletion with the Ribo-Zero kit, I don't share others' enthusiasm for ribosomal depletion versus poly-A enrichment. Yes, with ribosomal depletion, one can get all the non-codings RNAs without poly-A tails, and it may be more appropriate for degraded RNA. However, one also ends up with a lot of "junk", even when the quality of the RNA is good. I find the data from poly-A enrichment "cleaner", and I would recommend it over ribosomal depletion if one is not specifically interested in non-coding RNAs without poly-A tails.

I am familiar with the dilemma with human clinical samples. If the samples are unique and rare, the temptation is great to sequence them even if the quality is low. However, in my experience, trying to make sense of such data is a frustrating exercise.

--------------------------------------------------------------------

so how would you perform RNAseq on bacterial samples from human? Poly-A selection is not always an option....