Transcript coverage estimates?

[tboothby]

Hi all,

I have recently made de novo assemblies for the transcriptome of a plant microspore. The microspore is transcriptionally silent, so there is no new input or RNA. Basically the microspore stores all the transcripts it will need for its development and fertilization.

It is believed that this set of stored RNA is transcribed from only a very small subset of genes. If this is the case, then calculating coverage against the genome would give quite bad results (I would image).

Can any one comment on the appropriateness of estimating the coverage generated by Mbases of reads vs. Mbases of assembled sequence?

Would this even be an acceptable/informative value?

Basically I have 2145 Mbases of sequenced reads, that were assembled into 3.5 Mbases of 'unigenes.' I feel like this implies that each unigene on average should have a very large coverage ~600X but I don't know if this is a good way to look at things, or if I am just tricking myself into thinking the sequencing came out well.

Does anyone else have any suggestions for another approach?

[westerman]

As for the overall question, it is legitimate to say that you have up-to-X-coverage based on the total unigene size instead of basing it on the genome size. If you had the size of the total transcriptome (e.g., what you sequenced) you would use that but lacking that information using the unigenes -- which presumably encompass most of the transcriptome -- is a good starting point.


On a more specific level, you should be able to figure out exactly how many reads assembled into each unigene. That would give you a better feel (e.g., via a graph) of the coverage. In other words you may find that a handful of unigenes encompass most of the reads thus making many of the unigenes low coverage. Also you say that 2145 MBases of reads are assembled into 3.5 Mbases but it is unclear from that if all 2145 Mbases were used in the assembly or only a portion thus lowering the average 600x number.

[JingzhX]

i think you should use terms like FKPM OR RPKM to estimate the coverage of each unigene. As to the overall coverage, you should map all the reads to unigenes sets you get from assambly and then use the same principle of RPKM to evaluate the coverage.