de novo transcriptome

Question

[valopes]

Hi all,

I have 8 RNA-seq samples from the same organism in different conditions. After trimming, filtering quality, filtering possible contaminant species, and rRNA sequences, and et cetera, I've merged this high-quality reads and got a file with 98 million reads. From this, I've run Trinity and got ~4 million contigs. I found it a very high number! Anyway, let's go ahead with my question. I've found some contigs with ~200kb. When I check it for genes I can find something like 20-25 predicted complete genes. I was not expecting this. I mean, I don't know, is it something good or bad?

Thank!

[JC]

Are you working with a bacteria of virus? yes, that is expected as Operons. If not you maybe are creating chimeras in the assembly, how long are your reads? did you have paired-ends? which seq tech?

[valopes]

Okay! Sorry I miss a lot of info... It is Eukaryote, so not expecting operons. It is 2x150 pb Illumina.

Answer 1

[colindaven]

4m transcripts is a lot. I guess they are very highly redundant.

Filter - exclude very short ones.

You can reduce redundancy using a fasta clustering tool, eg cd-hit.

Gmap using GFF3 output is excellent for mapping to the genome and visualization.

Once you're checking the alignments visually you'll know how decent the quality is.

Good luck.

[valopes]

I will follow your suggestions. Thank you very much.