RNA-Polymerase III transcribed genes

Question

I have a list of differentially regulated genes from a high-throughput experiment of the human genome. I would like to find out which of these genes are transcribed by a RNA polymerase III and which of them (probably most) are transcribed by the RNA polymerase II.

I have downloaded some Chip-Seq tracks for polymerase II and III and some GO categories which apply for the RNA polymerase III transcription.

I have also tried the biomaRt repository to search for these mentioned GO categories, but I can't really find substantial results. For example I have the gene RN7SL1 which according to the literature is transcribed by PolIII, but I'm not able to find any repository which also says so.

Is there any (better) way to search for this kind of data/annotations?

thanks a lot Assa

PS. I am not sure, if I have found the correct GO categories (e.g. GO:0006383 - transcription from RNA polymerase III promoter). So are there any better one?

Answer 1

GO is based on experimental evidences; biomart is quite good. GO:0006383 seems to be the right tag.

If you want to use the ChIPseq data then you can do this:

• Map the ChIPseq reads to the genome. If you already have the tracks then you need not do the mapping. You would have the position of the reads in the genome.
• Now count the number of reads mapping to each geneic region. The gene boundaries can be obtained from a GTF file.
• Optional: Normalize the reads with the length of the gene.
• Find the difference between the number of Pol-II and Pol-III reads for each gene.

By doing this you still won't be able to say that "gene-x is transcribed by pol-III". All you can say is that "pol-III binds to that geneic region". If there is an even distribution of the ChIPseq reads along the gene then you can assume that there is an active transcription and the polymerase is not merely stalled. You can also do a CLIP (Crosslink-Immunoprecipitation) of Pol-III followed by a RNAseq. This will enable you to find RNAs transcribed by pol-III (this is a better way than ChIPseq but I don't think anyone has done it yet; so no ready data).