RNAseq quantification - Velvet assembler

[seq4franken]

Hi

we are using the Velvet assembler plugin in Geneious Server and were able to generate ca. 50,000 contigs from ca. 120 Mio paired reads (Illumina).

Unfortunately, the Velvet Geneious plugin does no longer produce the .afg file, which should contain all the informations on how many individual reads were used to assemble each contig.

I'm wondering whether there is a tool available which could map individual reads to contigs to find out which 'transcript/contig' was abundabtly expressed and which was a rare transcript. In my project, a rough estimation would be sufficient, and a comparision with other RNAseq runs is not necessary (I thought about re-blasting each contig to the reads, but this is quite time-consuming and extraction the no. of hits is not that easy).

Unfortunately, my Linux knowledge is limited, Windows would be preferred :-(

Any suggestion out there?
Thank you for suggestions!

[Torst]

There is an "abundance" of tools to align reads to contigs (pun intended).

My concern is that Velvet was not designed to assemble transcripts. It's underlying model assumes uniform abundance of all contigs. The "transcriptome" version of Velvet is called Oases. Is there a Geneious plugin for that?

[seq4franken]

Thanks for your post. I realized that there are 'abundant' tools available and I read the description of most of them. Unfortunately, I'm not an Linux expert, so most of them are not really suited to me. Thanks to Geneious that they make a Linux-bases tool (Velvet) available to 'average' users through their server plugin.

Unfortunaltely, there is no oases plugin available for Geneious Server. However, I realised the Oases has been developed by the same group and I will propose to develop a Plugin for Geneious. Should be quite easy since for their developers since they already did it for Velvet. I will see, how I can use it or whether eXpress (Berkely, mentioned in the Oases manual) could help me to get expression levels (I will give it a try with the contigs and the two paired reads; there is some sort of Windows support).

Thanks again for your valuable comments

[AdrianP]

I would like to help you, since I understand what your situation might be like.

I am not sure about the "server" version of geneious, but regular geneious allows you to allign reads to contigs. That will be a very resource intensive process, but if that is what you want to do, do it. Use the "Map to refrence" feature in geneious.

If you do have a Linux server, I can try and give you a list of commands that you can execute to get your assembly done with .afg files. But I will need the parameter info such as the k-mer you used, the exp coverage, everything you told Geneious.

[Jeremy]

You can generally do pretty well getting these programs to run with almost 0 linux knowledge. Just read the manual and type in the appropriate commands.
Or you can try the online version of galaxy, that is also very easy to use and has bowtie and BWA for mapping.

[seq4franken]

Hello,

thanks a lot for your replies.

@AdrianP: Geneious Server is also from Biomatters (who makes Geneious); more details are on their website. Our server has 192 GB RAM and 24 CPUs. So, we can handle quite large files on that server directly from the Genious client.
Since Oases is currently not supported by Geneious, I assembled paired RNAseq reads using Velvet/Oases on Linux command line:
a) Initial Velvet was done from 27 to 37 (as proposed by the Geneious Velvet Plugin, but run on Linux command line);
b) Velvet on transcripts was done with kmer 37 and merged with Oases

What I want to know now is how many original reads map back to the Oases assemblies.

@Jeremy: You are right, I should not be afraid too much by Linux. Once Velvet and Oases were installed (I neede help there), I coud do it myself.
Thanks for the Galaxy web service, I will have a look at it.

Best regards

[AdrianP]

Since you mention Geneious Server issues command lines, can you post its commands?

You mentioned you want an .afg file, right? That's you biggest issue so to speak.
If you know what command Geneious issues, and maybe if you can either modify it or issue it yourself, all you need according to:

http://www.ebi.ac.uk/~zerbino/velvet/Manual.pdf

is to add "-amos_file yes" parameter to velvetg when you run it.

[seq4franken]

I run Velvet/Oases directly from command line successfully. I also created an .afg file. The issue with Geneious Server is that there is no Oases plugin available for Geneious (although a Velvet plugin is available, but the output files are redirected to Geneious database).

My question is to find out how many reads were used to create each single transcript (contig) in the Oases output files. I think this info is not available directly in the output files.

I tried to re-map the original reads to the Oases transcripts by Geneious tools/plugins (Bowtie, Geneious Assembler, Tophat, ...). The problem there is that remapping the reads back to individual transcripts (ca. 78,000) causes some memory problems even on my big server (the way how Genious tools present the results would allow ,e to directly see the number of reads remapped to the transcripts). Remapping it to an "artificial transcript file" (concatenating all transcripts to one single sequence file) worked better with these programs on my server, but I loose the information of how many reads were used for each transcript segment in the artificial long sequence.

[seq4freedom]

it doesn't seem like you have the proper versions running... what version of Velvet are you running and what type of assembler is Genious... is it BOWTIE?

[seq4freedom]

yeah, Galaxy with 24 cores & 192GB of RAM is really great... you can probably find linux & Galaxy help here on the forum too!