reads of different size - RNA SEQ

[paolo.kunder]

Dear All,
I have just received some RNA Seq data and I noticed that not all reads are at the size I expected: (75bp)
only 64 % are at the correct size,
what should I do? Should I remove all reads below 75?
It is a library selection problem?

Many thanks,
Paolo

[GenoMax]

If this is illumina data then it looks like they must have been pre-trimmed. If that is the case go ahead and use them.

If not, then is this ion data?

[mastal]

Is it Illumina data? It looks like your reads have already been trimmed to remove adapters, and they should be fine as they are. Ask your service provider what trimming has been done to the data.

[paolo.kunder]

Yes this is Illumina NextSeq500 data,
I am a bit confused, why they should be of different length after trimming?
I analyzed many RNA Seq data of Hiseq and they are all (100%) the same length, I also downloaded 26 Human Tissued form Encode with Hiseq and they are all (100%) the same length.

[paolo.kunder]

and moreover, why should I have more than 300'000 reads 32bp long,

I have demultiplexed the data, with the following adapters

index
TCCGGAGA
CGCTCATT
ATTACTCG
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
TCCGGAGA
CGCTCATT
ATTACTCG
TCCGGAGA
CGCTCATT
ATTACTCG

[GenoMax]

These may be bad libraries that had short inserts which results in read-through into adapter on the other end. Those would generally need to be trimmed.

[GenoMax]

and moreover, why should I have more than 300'000 reads 32bp long,

I have demultiplexed the data, with the following adapters

index
TCCGGAGA
CGCTCATT
ATTACTCG
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
ATTACTCG
TCCGGAGA
CGCTCATT
GAGATTCC
TCCGGAGA
CGCTCATT
ATTACTCG
TCCGGAGA
CGCTCATT
ATTACTCG

Demultiplexing with these barcodes should have nothing to do with the length of the reads.

[paolo.kunder]

so is definitely a bad library preparation?
I have to be sure before going to complain to the service!

[GenoMax]

If these reads were pre-trimmed then they likely represent bad libraries. This may not necessarily indicate bad library preparation since the original samples themselves may be bad. One would need to look at the QC for samples and then libraries before concluding either one (or both) are bad.

[paolo.kunder]

these may help you?

[GenoMax]

This is outside my sphere of expertise so someone from experimental side of things will need to verify it authoritatively but it looks like these libraries have short inserts. Remember the size of the fragments include illumina adapters (~80 bp). Shorter fragments tend to cluster more efficiently.