Hej Marcelo,

This is something I've observed in my own NGS data sets. Mostly the co-receptors you mentioned display higher expression values than the tuning ORs and IRs, but there are some exceptions. I did not address this in the manuscripts I am writing, but surely it must hint at post-transcriptional implications for the proteins involved. That's not a good answer, I know, but I am not really sure if these phenomena have been explored in greater detail than speculation. In terms of basic biochemistry, energy utilization, there should be good reason why cells would mass-produce tuning receptor transcripts in greater abundance than their co-receptors. Also, I am not sure if we can yet assume a 1 to 1 stoichiometry for the receptors and the co-receptors. That has been the operational premise, but I am not sure that it has not been thoroughly investigated.

Cheers, 

Will. 

Hej Will!

Thanks a lot for sharing your view on this! I have the same feeling, though I am not that experienced as you are with this topic. I guess we know better some few model insects and prevailing concepts are based on that knowledge. Then, things that do not fit will show up if we start looking at least studied insects. But I would like other experts sharing their point of view on this, as it feels these are consistent results that suggest a more complex system.

Great to have your feedback!

Cheers,

Marcelo

I agree Marcelo. It would be great to hear what others have to say. By the way, I noticed a few typos or errors in my original response, so I've edited that a bit. 

-WW

Hej Marcelo, I would also suggest that we don't know that there is a 1:1 correlation between co-receptors and the unique receptors. For all we know at this time it could be 3 unique receptors to each single co-receptor... Something to think about.

Cheers,

Sharon

Hi Sharon,

thanks for your reply too! It would be really good to improve our understanding on these concepts. In fact it is fundamental to know more about how the molecules at the base of the system work and, how variable this OR/IR scenario can be.

Cheers,

Marcelo

I don't know the situation in insects, but depending on how many IR/OR genes you have and how similar they are (at the sequence level), you could have problems with aligning your NGS data on the genome for those genes and then get biased read counts... But it's maybe not the case, I don't know how many OR genes there are in insects...

Marcelo,

Another formal possibility is that the "tuning" receptors in those cases do not utilize the presumed coreceptors for complex formation. In any case, I wouldn't be at all confident in suggesting a particular subunit stoichiometry based on transcriptomics data. Moreover I think that Fabrice's comment is an excellent one. A lot of factors can contribute to the accuracy of read mapping in NGS data that may skew the output of expression values. For example, there can be 3' bias in library preparation, which may impact FPKM calculation if the 3' end of a particular receptor is not well annotated. One thing I like to do is use mapped sequence visualization software (like Seqmonk) to look for such artifacts in mapping data. Reannotation of gene models is sometimes necessary, followed by remapping to return more accurate results.

Fabrice, 

thanks for the input! Insects vary quite a lot in the number of OR/IR genes depending of the order. We know of some having as little as around 10 ORs (human lice) to several hundreds (flour beetle). I work with blood-sucking bugs that according to their genome sequence seem to have 110 OR genes (excluding 5 pseudogenes) and 33 IRs (IR numbers also vary from few to hundreds in diverse insect orders). 

I agree that there is a possibility that some read counts might have some bias due to extreme sequence similarity between genes, but as far as we have seen, some high expression results seem to be consistent through qPCR validation also. These results seem a bit shocking, considering our "established" concept of functional ORs being complexes formed by one OrCO unit plus one "tuning" receptor. Roughly speaking, the amount of OrCo transcripts should be similar to the sum of transcripts of all other ORs, which does not seem to really be the case in our dataset.

Ronald,

thanks to you too! I completely agree that transcriptome data cannot be used for estimating the way these receptors form functional units. Still, as I said above, I fear that the results I refer to suggest some conflicting proportions (OrCo vs ORs) that can be a good cue to look for in future studies. 

We indeed have re-annotated several of our target genes to correct the genome based sequences after comparing them with those from our RNASeq assemblies. Then we re-runned TopHat and still got some quite strikingly high expression in few specific ORs.

I confess that we have not looked at mapped sequences with such a software. Is it straightforward to use? 

Marcelo,

Seqmonk is freely available here (http://www.bioinformatics.babraham.ac.uk/projects/seqmonk/) and is straightforward to use. The An. gambiae genome is easy to install and you only need to import a mapped data file (BAM/SAM/etc.) to visualize reads against the annotated genome. Quite useful in many ways!

Cheers, Jason

Thanks a lot Jason!

We will try soon with our triatomine genome and reads to get a clearer picture of these things!

Cheers,

Marcelo

Boa Sorte!