Using Trinity package

[aevgup]

Hi everyone

I have some doubts about how to make trinity work for me.

I am currently working on RNA seq data from the Illumina. The reads are single end.
I am interested in 80 Ribosomal protein genes. I want to do transcriptome assembly for my 80 ribosomal protein genes. I came across this TRINITY package. It does de novo assembly.

Now here are the command line options i used for running my single end data using trinity

Trinity.pl --seqType fq --CPU 4 --single ERR030898.fq

This will generate several output files, 2 of which are single.fa and Trinity.fa.

Can someone explain me what are single .fa and Trinity.fa


After getting trinity.fa step what analysis should i do so that i get transcriptome assembly for only my ribosomal protein genes(80).

[westerman]

Suggest reading the downstream analysis part of Trinity.

http://trinityrnaseq.sourceforge.net/#Downstream_analyses

As for 'Trinity.fa' -- those are your assembled transcripts. In other words this is the file you want. 'single.fa' is an intermediate file generated from your initial data. You can ignore it.

[aevgup]

Trinity

Hey Rick
Thanks for replying. I am doing the downstream analysis of Trinity now, but still what i want to ask is how would this downstream analysis give me result of my 80 ribosomal protein genes i am interested in because Trinity.fa file have transcripts for everything , how to get the results for my interest i.e ribosomal protein genes.

Regards
Varun

[westerman]

Not to be too of a wisecrack here, but what do you already know about your 80 ribosomal proteins? Use that information to pull out your desired transcripts in Trinity.fasta.

In other words if you know the sequences of your proteins then you can use blast, blat, etc. to match the transcripts. If you know only that they are ribosomal then blasting the transcripts against 'nr' will give you keywords. Perhaps using a tool like blast2go will give even better refinement. Or perhaps you know something else about those 80 proteins that will help in the selection -- CG%? Length? But the point here is that only you know what information you have that can differentiate between ribosomal proteins and all of the other transcripts your organism. I can not help much aside from giving general ideas.

To use an analogy, say you are in a physical library full of books. You want all 80 books in the library which were written by Issac Asimov. How would you go about finding those books? You could look at the covers -- maybe his name is written on them. You could look inside the books in order to see if they are written in his style. Maybe you know that all of his books have a pink stripe on. Or perhaps you know nothing about his books and thus will have to wade through all of the books making an intelligent decision as to which books you want.

[aevgup]

Hey Rick
Thanks for the reply.
The only information which i have about my 80 ribosomal protein genes is genomic sequences for each gene combined in a single fasta format.

Can i use blast/blat to match with transcrips?

Regards
Varun

[westerman]

Hey Rick
Thanks for the reply.
The only information which i have about my 80 ribosomal protein genes is genomic sequences for each gene combined in a single fasta format.

Can i use blast/blat to match with transcrips?

Yes you can. It will be interesting to see what you pull out.

While Trinity is nice for de-novo work, when I have known reference then I use a program such as Tophat which can map reads versus the known. This is much more accurate (assuming a closely related and quality reference) than relying on de-novo methods.

[aevgup]

Hi
I used Tophat since i had the refernce but somehow the junctions.bed file produced was giving some weird results and so i thought of trying trinity

[westerman]

Trinity may certainly give you interesting results. For example, it may be entirely possible that you do not have sequence results for your organism. Perhaps you sequenced some strange fungus. It happens. By doing de-novo work you can find this out. Good luck!