Raw reads (from NGS) to perform a functional annotation analysis?

[Hel]

Hi all,

I have a theoretical question regarding functional annotation:

Is it correct to use raw reads (from NGS) to perform a functional annotation analysis?

Someone suggested me to use raw reads, but I think this idea has no sense, as they are incomplete genes. I have many sequences from RNA-seq and I think contigs are better, as they represent "genes". However, sometimes they match more than one gene…

I will appreciate your opinion,

Thanks in advance,

[dpryan]

Trying to annotate 10s or 100s of millions of individual reads wouldn't seem to be a very practical (let alone effective) endeavour. Depending on the species you're investigating, either de-novo assembling things or using a reference-based assembly strategy (I think those exist, this isn't something I ever need to personally do) would seem a better strategy.