metatranscriptomic read mapping

Question

[bsp017]

I mapped a depleted metatranscriptome bam file against a reference bacterial genome. Some genes have a 'substantial' number of reads mapping. I have previously isolated this bacteria from the sequencing site. Is there a method where I could say that x number of genes from this bacteria are expressed, therefore this bacteria is functionally active within the microbiome?

I have looked at the mapping data in Artemis and some genes are expressed >1000 times. How may reads constitute evidence of expression in a metatranscriptome?

Thanks

[Brian Bushnell]

Whether something is significant also depends on the denominator (total number of mapped reads). If you map 2 trillion Illumina reads, and generate conclusions based on a study that used 2000 Sanger reads, using the same absolute cutoffs, your results will be wrong.

Even ignoring that... I think it's much more important to do things because they make sense in the context of your experiment, than because some random unrelated paper happened to do that thing and got through the review process. I consider that to be bad science.

Answer 1

[bsp017]

In response to my question, I found this paper; which maps metatranscriptome files against an antibiotic resistance gene database. If reads map more than 3 times to a gene and span > 200 bp of that gene then it's considered significant.