RNA-seq assembly

[xy6699]

Hi all,

I'm trying to find some novel transcripts and estimate their abundances using RNA-seq data (Illumina GA II, 76bp, paird-end). I have tried tophat+cufflinks, but no interesting novel transcript was found, so now I want to try the de novo assemblers. I have learned there are Trans-ABySS, Trinity.....but not sure how well they work or whether they are suitable for my data.

Is there any suggestions for the software I shall use or any comments?

Many thanks

[jjohnson]

I personally like Trans-ABySS which analyzes ABySS-assembled contigs. But, you can;t go wrong with SW out of the Broad either. It may be best to try both.

But this may depend on the organism and amount of data you have. Trinity requires a GB of RAM per 1M reads. For Abyss, the single-processor version is for assembling genomes up to 100 Mb in size. The parallel version is implemented using MPI and is can assemble larger genomes.

[westerman]

The newer version of Trinity is not so memory intensive. I recently ran a 400M read assembly using about 140 GB memory. I am not sure if you can then say Trinity takes 140/400 GB per 1M reads but also it is obvious that the old rule of thumb (1 GB per 1M reads) no longer holds.

[xy6699]

Thanks a lot! It's human RNA-seq data and I have 20 samples (~1G per sample). Is it unrealistic to do the de novo assembly? Is the parallel version of Trans-ABySS capable to deal with human transcriptome?

[pbluescript]

RNA-seq, also called "Whole Transcriptome Shotgun Sequencing" [1] ("WTSS") and dubbed "a revolutionary tool for transcriptomics",[2] refers to the use of high-throughput sequencing technologies to sequence cDNA in order to get information about a sample's RNA content, a technique that is quickly becoming invaluable in the study of diseases like cancer.[3] Thanks to the deep coverage and base level resolution provided by next-generation sequencing instruments, RNA-seq provides researchers with efficient ways to measure transcriptome data experimentally, allowing them to get information such as how different alleles of a gene are expressed, detect post-transcriptional mutations or identify gene fusions.

If you're going to make such an obvious copy/paste, you should at least cite the source.

[gringer]

Looks like Wikipedia, based on a google search:

http://en.wikipedia.org/wiki/RNA-Seq

But regardless, that comment doesn't seem to add anything to the discussion of this thread.

If you want to use Trinity, the best approach is to pool your samples together and assemble using the pooled samples. It is possible with minimal effort to tweak the current version of Trinity so that it will run with your samples in under 100GB of memory (and most likely half that).