Is the RNA-seq library indeed strand-specific?

Question

[JJ]

Hi all,

So I am using RSeQC infer_experiment.py to check the strand-specificity of a public RNA-seq experiment. In the corresponding paper, the authors explicitly state that they have used the TruSeq™ Stranded Total RNA kit protocol. But this is what the output of RSeQC infer_experiment.py looks like:

This is PairEnd Data
Fraction of reads failed to determine: 0.0787
Fraction of reads explained by "1++,1--,2+-,2-+": 0.4686
Fraction of reads explained by "1+-,1-+,2++,2--": 0.4527
Finshed Checking Strand Specificty

So not strand-specific ... Another Run of the same experiment looks a bit better

This is PairEnd Data
Fraction of reads failed to determine: 0.0946
Fraction of reads explained by "1++,1--,2+-,2-+": 0.6052
Fraction of reads explained by "1+-,1-+,2++,2--": 0.3002
Finshed Checking Strand Specificty

Usually I get something like this:

This is PairEnd Data
Fraction of reads failed to determine: 0.0500
Fraction of reads explained by "1++,1--,2+-,2-+": 0.0287
Fraction of reads explained by "1+-,1-+,2++,2--": 0.9213
Finshed Checking Strand Specificty

So I suppose the stranded protocol did not really work for the data set in question... However, my question now is: Is there a guide for the fractions? what fraction is still ok for the data to be strand-specific? Is there a cutoff? I sometimes get for Fraction of reads explained by "1+-,1-+,2++,2--": 0.7 / 0.8 - is this still ok?

Thanks for your input!

Answer 1

[Devon Ryan]

There's no strict guide for this, it's all just eye-balling the numbers. But as you correctly determined, either they didn't actually use a strand-specific protocol or they have a LOT of anti-sense transcription. In my opinion, if the ratio is worse than 0.9/0.1 and you don't expect a lot of anti-sense transcription then there are likely other major issues with the dataset. At that point you have to consider whether it makes sense to proceed at all with the analysis.

[JJ]

Thanks for the answer.

Did you get something like this before?

This is PairEnd Data
Fraction of reads failed to determine: 0.9938
Fraction of reads explained by "1++,1--,2+-,2-+": 0.0003
Fraction of reads explained by "1+-,1-+,2++,2--": 0.0059
Finshed Checking Strand Specificty

Not sure what to make of this...

[Friederike]

interesting...what organism is this? did you notice any other problems with this data set, e.g. after running FastQC?

QoRTs also has a script that checks strand-specificity, would be interesting to see what it spits out for those data sets.

[JJ]

No, I get something similar from an Encode bam file I downloaded.

Fraction of reads failed to determine: 0.7454
Fraction of reads explained by "1++,1--,2+-,2-+": 0.0039
Fraction of reads explained by "1+-,1-+,2++,2--": 0.2506

https://www.encodeproject.org/experiments/ENCSR274JRR/ ENCFF978EVS.bam

thanks for the link to that tool! I'll try it out :)

oh, and it's human data

[JJ]

Could an insert size of 0 be an issue for the tool? This is the only communality I found between getting "weird" output and the different public datasets, although I don't know about the Encode one - I couldn't find an insert size for this one.

[Devon Ryan]

That's a first, I presume there's chromosome naming issue or something.

[JJ]

So I did

infer_experiment.py -r Homo_sapiens.GRCh38.79.bed -I ENCFF978EVS.bam

got this

Reading reference gene model ../../Homo_sapiens.GRCh38.79.bed ... Done
Loading SAM/BAM file ...  Total 200000 usable reads were sampled


This is PairEnd Data
Fraction of reads failed to determine: 0.7454
Fraction of reads explained by "1++,1--,2+-,2-+": 0.0039
Fraction of reads explained by "1+-,1-+,2++,2--": 0.2506

[Devon Ryan]

Maybe the BAM file is aligned against hg19? Have a look at the lengths of a couple chromosomes.

[JJ]

Well, I selected GRCh38 ... It's a different version though GRCh38 V24.

[JJ]

hm, I don't think so - I check this beforehand and edited the bed file...

samtools idxstats ENCFF978EVS.bam | cut -f 1 | head
chr1
chr2
chr3
chr4
chr5
chr6
chr7
chr8
chr9
chr10

and

cat Homo_sapiens.GRCh38.79.bed | cut -f 1 | uniq
chr1
chr10
chr11
chr12
chr13
chr14
chr15
chr16
chr17
chr18
chr19
chr2
chr20
chr21
chr22
chr3
chr4
chr5
chr6
chr7
chr8
chr9
chrCHR_HG126_PATCH
chrCHR_HG1362_PATCH
chrCHR_HG142_HG150_NOVEL_TEST
chrCHR_HG151_NOVEL_TEST
chrCHR_HG1832_PATCH
......