De novo transcriptome assembly using Trinity but for single-end reads?

[Sebastian_Quezada_R]

Hi everyone =)

A colleague of mine advised me to create a de novo transcriptome assembly for my reads in sheep (because sheep annotations suck) but we couldn't find a way on the internet to do it for single-end reads. All of them were for paired-end and started by aligning Fw and Rv reads, so we're not quite sure how we should do it.

Have you done it before?

Cheers,

Seb.

[gringer]

Trinity will work with single-end reads (use argument '--single'), it's just a bit better with paired-end, and even better with stranded reads. Paired ends can help to resolve different isoforms, and there are some instances where transcripts can overlap on opposite strands.

[kmcarr]

Before going to the trouble of doing your own transcriptome assembly I'd test just how much the current annotations suck. The most recent Ensembl genebuild for sheep has ~21,000 coding genes and ~6,000 non coding. Map your read data to the genome and see what percentage of your reads map within annotated genes. Also, since a genome sequence exists you may consider using a reference guided transcriptome assembly instead of de novo.

[Sebastian_Quezada_R]

Trinity will work with single-end reads (use argument '--single'), it's just a bit better with paired-end, and even better with stranded reads. Paired ends can help to resolve different isoforms, and there are some instances where transcripts can overlap on opposite strands.

Thanks =) That worked awesome

Before going to the trouble of doing your own transcriptome assembly I'd test just how much the current annotations suck. The most recent Ensembl genebuild for sheep has ~21,000 coding genes and ~6,000 non coding. Map your read data to the genome and see what percentage of your reads map within annotated genes. Also, since a genome sequence exists you may consider using a reference guided transcriptome assembly instead of de novo.

Which program would you recommend to be the best for that? I might do the reference guided transcriptome assembly afterwards if this doesn't work out.

Thanks in advance!

[gringer]

Reference-guided Trinity for sheep should be fine (see here), and would improve your transcriptome assembly. The main issue with that is that you won't see any reads that aren't supported by the genome, but the sheep assembly should be comprehensive enough that you shouldn't have any problems. At least, that's the impression I got from James Kijas today when he was talking about it at QRW.

[Sebastian_Quezada_R]

Reference-guided Trinity for sheep should be fine (see here), and would improve your transcriptome assembly. The main issue with that is that you won't see any reads that aren't supported by the genome, but the sheep assembly should be comprehensive enough that you shouldn't have any problems. At least, that's the impression I got from James Kijas today when he was talking about it at QRW.

Thanks for that =)!