If you don't have a reference genome, you can't predict the genes. Gene prediction is the process to identify genomic DNA regions that encode genes (http://en.wikipedia.org/wiki/Gene_prediction). So, you don't need it.

Still you may need to identify transcripts that encode proteins for your analysis removing the non-coding RNA (http://en.wikipedia.org/wiki/Non-coding_RNA), in which case you can use two approaches. a) Sequence homology search of the translated transcript sequence with a protein database such as GenBank NR or TREMBL using BlastX. b) ORF predictions tools such as ESTScan or TRansdecoder (from Trinity package http://trinityrnaseq.github.io/analysis/extract_proteins_from_trinity_transcripts.html)

Hope this helps,

thank u for the reply its really helpful

Hi Lakshmi,

If you already have the annotation, I think that would be more than enough. Unless you are analyzing gene structure (exons and introns) then the annotation would give you the product of the gene which is probably what you are looking for.

If you still want to have the genes, you will need to go for DNA sequencing or try to find in your RNAseq if you have some unspliced transcripts, but that would be a very long shot and you will have to reconstruct them. This is an idea I've been working on for a while but so far, not very good results...

Good luck.

thanks for the reply

I went through one of your publication mentioned below got some doubts like how to correlate the transcriptome sequences to that of genes do you have any automated pipelines for the annotation?

Mango (Mangifera indica L.) cv. Kent fruit mesocarp de novo transcriptome assembly identifies gene families important for ripening

Hi Lakshmi

For the mango transcriptome we used trinity pipeline for transcript reconstruction. The pipeline assemble the reads and solve isoforms from the debruijn graphs. Each crap component is assumed to be a gene that can generate 1 or more transcripts. This is how the program gives you an estimate of how many genes you have but it never gives you the gene structure since it can predict the splice sites or introns.

For the annotation we use trinotate which basically integrates blastx, blastp, pfam, signalp and tmhmm searches. Blast searches are performed on swissprot and uniref databases. For pfam, pfamA database is used.

Hope this will answer your question.

yes sir thank you

In my work I have used iAssmebler since my data was generated using 454 roche pyrosequencing  I have used blastx, pfam , priam & GO for annotation and got a few new enzymes in the annotation. should i to go for signalP  and tmhmm before proceeding to pathway reconstruction.