Transcriptome assembly using solexa reads

[baohua100]

species 1 : 10M 35bp reads
species 2 : 12M 35bp reads

de nove assembly：
species 1: 4235 contigs
species 2: 5863 contigs
and we known the coverage of every site on each contig.

How can we use coverage information for gene expression analysis of one species or comparative analysis between 2 species?

[mchaisso]

Perhaps concatenate the two datasets into one, and then parse the reads mapped to each contig to determine the mix of reads mapped to the contig?

[jarret_glasscock]

Hello baohua100,

What did you use for your transcriptome assembly?
I have been using velvet, however the need to put in the average coverage makes it appropriate for genomic assembly but not transcriptome (where gene to gene expression levels vary greatly).

best.
jarret

[baohua100]

Hello baohua100,

What did you use for your transcriptome assembly?
I have been using velvet, however the need to put in the average coverage makes it appropriate for genomic assembly but not transcriptome (where gene to gene expression levels vary greatly).

best.
jarret

I want to see how solexa de nove transciptome sequencing to get ESTs of non-model plant. I use Edena to assembly.

[masterpiece]

I want to see how solexa de nove transciptome sequencing to get ESTs of non-model plant. I use Edena to assembly.

hi boahua,

I've found a paper published on transcriptome analysis using non-model plant. here s the url linked to the paper http://www.jsbi.org/modules/journal1....php/GI21.html

[masterpiece]

Hello baohua100,

What did you use for your transcriptome assembly?
I have been using velvet, however the need to put in the average coverage makes it appropriate for genomic assembly but not transcriptome (where gene to gene expression levels vary greatly).

best.
jarret

hi jarret

how's your assembly? currently i also doing assembly using velvet, but having trouble on selecting hash length that give the best assembly result. do you have any idea?

agreed with you about the average coverage is not appropriate for transcriptome analysis.

going for Edena to see how's assembly looks like.

[swbarnes2]

hi jarret

how's your assembly? currently i also doing assembly using velvet, but having trouble on selecting hash length that give the best assembly result. do you have any idea?

agreed with you about the average coverage is not appropriate for transcriptome analysis.

going for Edena to see how's assembly looks like.

What if you tried a few rounds of:

1) velvet up contigs with a pretty high depth threshold
2) pull out those sequences, or the transcripts they represent
3) align against those transcripts with a fairly low tolerance for errors
4) velvet the reads that that are leftover with different settings
5) pull out those transcripts align, etc

It's just an idea, but the tiering might allow you to apply different settings to contigs with differnt depth.

[baohua100]

hi jarret

how's your assembly? currently i also doing assembly using velvet, but having trouble on selecting hash length that give the best assembly result. do you have any idea?

agreed with you about the average coverage is not appropriate for transcriptome analysis.

going for Edena to see how's assembly looks like.

I am looking forword to hear your results from Edena.

[masterpiece]

What if you tried a few rounds of:

1) velvet up contigs with a pretty high depth threshold
2) pull out those sequences, or the transcripts they represent
3) align against those transcripts with a fairly low tolerance for errors
4) velvet the reads that that are leftover with different settings
5) pull out those transcripts align, etc

It's just an idea, but the tiering might allow you to apply different settings to contigs with differnt depth.

thanks 4 the idea,

correct me if I'm wrong. its possible to pull out contigs that have high depth threshold, but howbout those which are lowly trancript, I think those contigs will have low depth coverage.

[deks84]

Help me please..

I don't really get what all of you have been discussed here.

Recently I've being asked to do transcriptome assembly by my supervisor. But the thing is we don't have anyone here that can help me to do it. We only have those who know assemble genome data. I don't even have any idea of transcriptome assembly. Please tell me how to start, I mean which type of data I should work first? because we have outsourcing to do the RNA-seq for our samples. But in order to assemble that I don't know how. Please help me.

[strob]

http://www.ebi.ac.uk/~zerbino/oases/

De novo transcriptome assembler for very short reads (the RNAseq variant of velvet)

[deks84]

http://www.ebi.ac.uk/~zerbino/oases/

De novo transcriptome assembler for very short reads (the RNAseq variant of velvet)

Thank you for the info. May I know what do you mean by very short reads? is it only for prokaryotic? How about eukaryotic like human?

[Linnea]

"Very short reads" refers to the sequencing technique, not the organism. I would guess very short reads in this case are Illumina and SOLiD reads, up to ~75-100 bp.
Oases should be fine to use with eukaryotic data, but you might need a tremendous amount of RAM...

You could also use other software designed for assembling genomic data, like SOAP denovo or ABySS. ABySS has been used for assembling a human transcriptome (http://bioinformatics.oxfordjournals...ull/25/21/2872).
I'm trying both for assembly of a bird transcriptome, they seem to work (I'm not nearly finish though..)

[bioinfosm]

Hey that looks neat. Have you tried it at all? It would be interesting to compare it with tophat-cufflinks results of mapping to a reference!

What if you tried a few rounds of:

1) velvet up contigs with a pretty high depth threshold
2) pull out those sequences, or the transcripts they represent
3) align against those transcripts with a fairly low tolerance for errors
4) velvet the reads that that are leftover with different settings
5) pull out those transcripts align, etc

It's just an idea, but the tiering might allow you to apply different settings to contigs with differnt depth.