Paired-end RNASeq insert length for de novo assembly using Oases

Question

[Urja]

Hi Fellow Bioinformatician,

I am doing de novo RNASeq assembly using Velvet/Oases for a few publicly available RNASeq datasets. They are paired end 101bp reads generated by Illumina. I read and found that Oases has a parameter for insert length (-ins_length), but I have no idea how much the insert length is and I have following questions:

1. Can I run Oases on paired end RNASeq data without inputing an insert length ? And what would be the default behavior in that case ?

2. Is there any way to find out insert length for publicly available paired end RNASeq data if it's not provided in the description or article ?

Thanks for your help!

Urja

Answer 1

[WouterDeCoster]

Is a reference genome of your species of interest available? If so, you could just map the RNAseq reads to the genome and calculate the insert size based on the mapping distance of the read pairs...

Answer 2

[Urja]

That's a good idea but unfortunately, no reference genome is available.

I noticed that even without giving an insert size Oases runs and does an assembly, but I am not sure if it considers the paired information or just treats them as single end reads. Any thoughts on that ?

Thanks a lot for your help!

[GenoMax]

Do you know if the reads overlap (R1/R2) in any of these datasets. If that is true then you could get an estimate of the insert size for those datasets.
You could also align R1/R2 reads to the assembly you got to see if you are able to get them to align in a logical fashion (and estimate the insert distance).

Answer 3

[Urja]

Thank you. I will align reads to the assembled transcriptome and check for overlaps. Any idea how oases deals with paired end reads if no insert size is specified?

[GenoMax]

Please use the ADD REPLY/ADD COMMENT buttons when responding to previous posts. That way the responses stay with the relevant posts.

[GenoMax]

The reads themselves (R1/R2) should not overlap if these are standard illumina libraries (and the read lengths are shorter than 200 bp). You con confirm that using BBMerge from BBMap or FLASH or similar programs. Oases manual indicates that if the reads overlap then they be merged beforehand.
I don't know for sure about oases but it must use a default value for insert size if you don't specify one.