wheat transcriptomes

[Triticum]

hi!
we want to analyze 10 different wheat transcriptomes. we planned to use 454-sequencing, but i´m not sure if this is the methode of choice (because of the transkriptom-size, needed coverage and the resulting high price)

which methode would you use?

thanx for some input! Triticum

[karthikprabu]

which triticum sp.. u r studying ...

[Triticum]

some NILs of t. aestivum

[karthikprabu]

ya me 2 studying the wheat transcriptome

[krobison]

What questions are you trying to ask about these transcriptomes? The primary advantage of 454 is long read length -- are there questions for which this is critical?

Based on the conversations I've had with folks, SOLiD is probably the cheapest way to get a lot of data and Illumina with paired ends would give a bit less data/$ but give you some long-range information.

[Triticum]

great! concerning also resistance? which methodes do you use for sequencing and what are your experiences?

[karthikprabu]

i use illumina genome analyzer

[Triticum]

its about the involved genes in fusarium-resistance. so i think longer fragments are never bad but i don´t know if we really need them in that case. annotating genes should also work when assembling the shorter fragments (with probably a deeper coverage), isn´t it?

[karthikprabu]

for this u can go for RNA seq

[krobison]

First the disclaimer: I'm only an armchair next-gen sequencer -- still planning my first project. I've read a lot & spoken with a number of service providers, but that's it.

What you need to decide is what genomic question you are asking. I'll give four possible scenarios

1) If you are most interested in determining which transcripts are different between your strains AND you trust the existing wheat transcriptome database, then you should use SOLiD with a protocol which reads only the 3' end of the message. Reading only the end removes length-based biases and SOLiD will give you the most reads for the lowest cost, meaning the deepest exploration of the libraries and highest sensitivity for rarer transcripts.

2) If you are interested in finding expressed SNPs, then use another RNA-Seq method and either SOLiD or Illumina. SOLiD will still give the most data at lowest cost, but will be hard to assemble. Illumina can get much more data per fragment (by using paired ends plus the generally longer reads of Illumina). You'll still get some transcript abundance information, but with a bias towards longer messages

3) If you are most interested in novel splice forms and trust the existing transcriptome, then use Illumina with paired ends. This will be a sensitive way to work out various splice forms but will give a lot of data.

4) If you are most interested in novel transcript discovery -- i.e. you don't trust the existing transcriptome sequence information for your species, then 454 is probably best. Some people would choose Illumina with paired ends for this as well, to trade more data for a more difficult assembly problem

[Triticum]

thanks krobison for your post!
our aim is to compare 10 different transcriptomes. these transcriptomes are derived from 5 different near isogenic wheat-lines, all fusarium-inoculated and mock-inoculated (=> 10 samples). so we want to compare the transcriptomes and hope to find some transcripts which differ and are therefore probably involved in the resistance to fusarium.
for this, i think case 1) will suit best, but detecting SNPs is also intended to perform.
in 2), did you mean to do solid or illumnia AND another methode (eg 454)? using solid alone does not seem to be useful for detecting SNPs. istn´t the best method for SNPs using 454 because of the length?

[kmcarr]

our aim is to compare 10 different transcriptomes. these transcriptomes are derived from 5 different near isogenic wheat-lines, all fusarium-inoculated and mock-inoculated (=> 10 samples). so we want to compare the transcriptomes and hope to find some transcripts which differ and are therefore probably involved in the resistance to fusarium.
for this, i think case 1) will suit best

1) If you are most interested in determining which transcripts are different between your strains AND you trust the existing wheat transcriptome database, then you should use SOLiD with a protocol which reads only the 3' end of the message. Reading only the end removes length-based biases and SOLiD will give you the most reads for the lowest cost, meaning the deepest exploration of the libraries and highest sensitivity for rarer transcripts.

I don't think that you could use a sequence tag method (i.e. SAGE or DGE) for wheat. To use a method like this you need to already have in hand very well defined transcriptome. This is not the case for T. aestivum. At best there are currently a set of putative transcript assemblies (unigenes), most of which represent incomplete transcripts. Identifying SNPs with confidence will also be difficult; you will probably not have any information on the confidence of the base calls in the putative transcript assemblies used as a reference so you could not say for certain if differences observed are true SNPs or read errors in the sequencing of reference. Then there is the issue of paralogous transcripts. With T. aestivum having a hexaploid genome it will be difficult to know for sure if an observed base variation is due to a true change in an allele or if you are aligning a read from a paralogous gene.

Given what is currently available to use as a reference I would probably go with either Illumina or SOLiD (whichever is most available to you) short reads (36nt) and then map these reads to a reference set of either wheat Unigenes or the TIGR Plant Transcript Assemblies for wheat. These sets are bound to be highly redundant and you will get a lot of non-unique matches. It is likely you will not be able to identify specific genes but more likely classes of genes which are differentially expressed.

I know that this is a sequencing forum but have you considered using microarrays? Agilent has a wheat gene expression array available with ~44,000 features. I don't know the cost but it may be significantly cheaper than sequencing, especially if you wish to analyze many samples.

[westerman]

2), did you mean to do solid or illumnia AND another methode (eg 454)? using solid alone does not seem to be useful for detecting SNPs. istn´t the best method for SNPs using 454 because of the length?

krobison undoubtedly meant exact that -- use the SOLiD or the Illumina for SNP detection -- if you trust your reference sequence(s). If you have a good reference and thus you can map reads to the reference then increased length does nothing for SNP detection. What is better is to have a large depth of coverage.

Another way to think of this is asking yourself, "what is the length of a SNP versus the length of a read?" Obviously the length of a SNP is "1 base". This small of a change can easily be detected by reads of 25 bases -- the more the better. This is where the Illumina and SOLiD excel. The SOLiD is particularly good at SNP detection because its "color-space" encoding system means that one can differentiate between single base sequencing errors and true SNPs.

Now if you are looking for InDels or rearrangements then longer reads (especially paired reads) are better. And if you don't trust your reference and thus have to create a de-novo reference then, like any de-novo assembly project, longer reads are better.

I once did a project similar to yours where we were looking for transcriptome counts but did not trust our reference. Thus we pooled all of the samples and ran them on a 454 in order to create contigs. Then we ran the individual samples on a SOLiD in order to get the counts. This seemed to work well enough.

But if I had a good reference to begin with I would never run the sample on a 454 just to get SNPs. Not with the Illumina or SOLiD being available.

[Triticum]

@kmcarr: we already performed a microarray - sequencing is the next step.
@all: thanks for your advises. seems that solid would be a good choise, but of course there remains the problem that there is no good reference for wheat. pretty hard...

thanks. triticum.

[bioenvisage]

reg SNPs in wheat

hi ..R u using the inbreds to find the SNP?