Genome guided vs de novo - Trinity

[Mocca]

Hi everyone,

For a while now I have been working on my de novo Trinity assembly (fish species). Using my de novo transcriptome I have been able to do some interesting differential expression analyses. Then the genome-guided option of Trinity was launched and I tried it using the second version of our draft genome. I didn't use the first genome version as some of my genes of interest were poorly (wrong) assembled and also, on occasion, partially within gaps. Now this is mostly corrected and I hoped that doing the genome guided Trinity would reduce the number of transcripts. I went from 320520 'genes' and N50 1235 to 342099 'genes' and N50 1716.

So better N50 but more transcripts... Also, if I do abundance estimation a FPKM cutoff of 2 for the first gives ~30 000 'genes' and the second ~119 000 'genes'.

Based on stats only - which would you chose to use for DE analysis? Considering redoing Trinotate and all DE analyses with the genome guided version.

Any thoughts are most welcome

[gringer]

Genome guided should be better with a good reference genome. The main disadvantage is that transcripts will only be generated if they are supported by the genome, so if the reference genome assembly is incomplete then there might be a few missed transcripts.

[Mocca]

Discussing the quality/"completeness" of the genome with my collaborators migth be the next step then. I also got a bit hung up on the abundance estimation analysis with the FPKM 2 cutoff where the number of transcripts suddenly fit the number of predicted genes for my species using the de novo assembly but not the genome guided assembly. Could just be coincidences though.

If I had a complete list of "genes of interest" I could check if they were assembled properly in the genome/transcriptome, but I'm also searching for alternative responses to treatment so...

Contemplating a bit more :-)