Trinity assemblies do not contain isoforms?!

[michaelsbrewer]

I recently assembled a number of transcriptomes using the Trinity pipeline. All proceeded smoothly until I began looking for isoforms. Most of the assemblies do not have any while a few have only a hand full (~5). All of the comp#s end in "_c0_seq1". My sequencing was not terribly deep; we multiplexed four individuals per lane of an illumina flow cell for RNAseq. Perhaps the sequencing was not deep enough to recover many isoforms?

I guess what I am asking is, how many isoforms should one expect from a Trinity assembly (e.g., 22,623 sequences in the Trinity.fasta output; average length of 610 nucleotides) sequenced in such a manner?

Thanks,
Michael

[Jeremy]

Number of isoforms depends on your sample, for example bacteria would not have many whereas a human would have more. Also, is it paired end data or single end, paired end data has more information that can be used to identify isoforms than single end.

[michaelsbrewer]

Jeremy,

The data were paired-end reads obtained from whole body extractions of arthropods (millipedes and spiders). Data concerning isoforms from these taxa are nearly non-existant as little work has been done in these organisms. Any advice or insight would be appreciated.

[Jeremy]

I guess either there just aren't many alternately spliced genes in millipedes and spiders or Trinity doesn't perform well with these organisms. As with all NGS data, try a few other assembly programs and see how they look.

[Apexy]

Given that trinity is locked in a single k-mer could be another reason why you have few isoforms. You should be able get diverse rnatigs with lower k-mers. Like Jeremy mentioned, it depends also on your data (false k-mers introduced by sequencing errors). Trying out another assembler is good, and you should envisage how you will be able to filter novelty (genuine isoforms) vs noise (bogus isoforms).