De-Novo transcriptome with 454 Titanium

[flxlex]

Being a bit focused on newbler, I would wait till the new release which includes a transcriptome module for assembly. It generates transcript variants (for example alternative splice variants) grouped according to gene.

To be fair, there are certainly other programs out there that might do a good, or even better job, for example Abyss: http://www.ncbi.nlm.nih.gov/pubmed/19528083 for Illumina reads (perhaps this works also for 454 reads?).

[yvan.wenger]

Yep, most of the publications are focused on "pre-titanium"
technologies and de-novo transcriptome applications are not so
common... But I bet that even if the scientific litterature is not so
prolific at this time, many people are currently working on it.

[krobison]

If you search PubMed for "454 AND transcriptome" it gets quite a haul; have you found any of those papers useful or are they too dated (a general problem with this field; the technology evolves faster than people can push publications out!)

[yvan.wenger]

Hello everybody,

Has somebody done transcriptome assemblies with 454 Titanium sequences?

The goal of my experiment would be to assemble a comprehensive "transcriptome space" and not to do anything quantitative. Based on preliminary experiments, I expect roughly 15'000-20'000 different genes representing between 25x10⁶ and 50x10⁶ Mb refseq-like transcripts (including all possible exons in each reference sequence).

I*would particularly be interested if someone could advise/speculate on one or several of these points:

- PolyA enriched samples vs Normalized samples (loss of rare transcripts? efficiency?)

- Paired end librairies (overlapping maybe? or particular insert sizes?) vs single end librairies

- The quantity of reads (or plates assuming 1x10⁶ reads per plate) required to signal a good fraction of all the transcripts expressed (e.g 95-98%)

- The "average" (for the whole sequencing)*or "minimum" (for rare transcripts) coverage required to obtain a good contiguity after de-novo assembly (e.g. 80%+ full length transcripts)

- The incidence of the points mentioned above on the presence of rare transcripts such as transcription factors in the final assembly.

Please don't hesitate to point to recent publications or other threads.

Cheers,

Yvan