Thursday, August 18, 2016

Do you believe what's written in the introduction to this paper?

I came across this paper while doing research on alternative splicing. The introduction annoyed me. It illustrates what to my mind are some serious problems with modern scholarship.

Scotti, M.M. and Swanson, M.S. (2016) RNA mis-splicing in disease. Nature Reviews Genetics 17:19–3 [doi: 10.1038/nrg.2015.3]
Here's part of the first paragraph in the paper.
Recent analysis from the Encyclopedia of DNA Elements (ENCODE) project (GRCh38, Ensembl79) indicates that most of the human genome is transcribed and consists of ~60,000 genes (~20,000 protein-coding genes, ~16,000 long non-coding RNAs (lncRNAs), ~10,000 small non-coding RNA and 14,000 pseudogenes). Although this gene inventory will change with further analysis, the number of protein-coding genes is surprisingly low given the proteomic complexity that is evident in many tissues, particularly the central nervous system (CNS). High resolution mass spectrometry studies have identified peptides encoded by most of these annotated genes, but the number of isoforms expressed from this gene set has been estimated to be at least 5–10-fold higher. For example, long-read sequence analysis of adult mouse prefrontal cortex neurexin (Nrxn) mRNAs indicates that only three Nrxn genes produce thousands of isoform variants. This diversity is primarily generated by alternative splicing, with >90% of human protein-coding genes producing multiple mRNA isoforms.
Here are some of the problems I have with this introduction. My opinions on these issues differ from those of the authors.
  1. I think that pseudogenes are not genes.
  2. I think there are NOT ~16,000 lncRNAs and ~10,000 small-noncoding RNA genes. Instead, there are approximately this many putative or predicted genes, many of which will undoubtedly turn out not to be genes. Some of them will be pseudogenes.
  3. I don't think there's a discrepancy between the known number of protein-coding genes and proteomic complexity; therefore, it is misleading to say that the number of protein-coding genes is "surprisingly low."
  4. I'm pretty sure that nobody has ever proposed a truly scientific "estimate" of isoforms showing that the number should be 5-10-fold higher than the number of genes. This is all speculation and guesswork based mostly on deflated egos.
  5. It is not true that >90% of human genes produce multiple mRNA isoforms by alternative splicing. What IS true is that for every human gene researchers have detected low levels of non-canonical splice events upon careful analysis of the transcriptome. We do not know whether these represent true biologically relevant alternative splicing or simply splicing errors. All available evidence suggests that the vast majority are splicing errors.
The authors are certainly entitled to their opinion ... even if it differs from mine!

But surely there has to be a better way of expressing this opinion to make it clear that they aren't stating facts but just their own personal views based on their own interpretation of the literature? This becomes very important if there's widespread scientific controversy over some of these opinions. (It's not so important if there's widespread agreement, or consensus, in the scientific community. In those cases, you aren't obliged to mentions alternative views held by kooks.)

I believe that scientists have an ethical obligation to distinguish between fact and opinion and to make it very clear in their writings which is which. I don't know whether Scotti and Swanson know about the controversial aspect of their statements and are deliberately avoiding any mention of them, or whether they actually believe that their statements are factual. Either way, we have a problem.


11 comments :

  1. FYI - Your link for the "deflated egos" post wasn't formatted correctedly. It can be accessed here:

    http://sandwalk.blogspot.ca/2007/05/deflated-ego-problem.html

    ReplyDelete
  2. I think I'd take issue with your claim #4, that the estimate of the number of isoforms is just "speculation and guesswork"
    My understanding of mass spec (based on reading a single paper years ago) is that they fragment the proteome and determine the sequence of each fragment. Its relatively straight-forward to determine which fragments dont correspond to any annotated variants and to determine which gene they arise from. From there one could get a reasonable number for average splice variants. I dont see any other way they could make that claim based on HRMS data. If its doesn't follow then they're making the claim despite the evidence

    ReplyDelete
    Replies
    1. I've read all the papers on mass spec analysis of the human proteome. They detect very few of the several hundred thousand protein variants predicted by splice variants. Even the ones that are detected could be nonfunctional peptides translated by accident from an incorrectly spliced RNA.

      I've never seen any evidence in any of the papers showing that the average human gene produces 5-10 different biologically functional isoforms.

      How many proteins do humans make?

      Here's what Kim et al. say in their paper ...

      Alternative splicing gives rise to a large number of splcie variants at the RNA level. Multiple protein isoforms are contributed by only one-third of annotated genes, whereas the remaining two-thirds generate only a single protein product according to the RefSeq database. Although our primary goal was not to obtain complete coverage of splice isoforms, we identified isoform-specific peptides for 2,891 protein isoforms for 2,450 genes.

      In other words, about 12% of human protein-coding genes were shown to produce multiple protein (peptide) isoforms and, of this subset, the vast majority produced only two different isoforms.

      BTW, the latest RefSeq database has eliminated almost all of the splice variants detected in earlier studies. They have concluded that most of them are artifacts or splicing errors. They are not biologically relevant. The remaining ones are not necessarily examples of true alternative splicing but they can't be conclusively ruled out.

      Kim, M-S. et al. (2014) A draft map of the human proteome. Nature, 509:575-581. doi: 10.1038/nature13302


      Delete
    2. Larry wrote:

      I think that pseudogenes are not genes
      What are they then?
      If I were to survey geneticists, what percent of them would agree with you?
      What evidence do you have that supports your view other than the usual, boring one?

      Delete
    3. Cruglers, pseudogenes ("false" genes) resemble functional genes, but have some mutation(s) that make them not functional. For example, primates (including you and me) have a pseudogene for an enzyme needed to make vitamin C. What we have clearly resembles the functional vitamin C gene in other animals, but most of the gene is missing in us. I don't know if it makes any protein fragment, but it certainly doesn't make a functional one.

      Because the pseudogene does not code for a functional protein, it isn't a gene now, though it's ancestors were genes, long long ago.

      Delete
    4. And humans, chimps and other apes have the same disabling mutations in the same places in their pseudogenes for vitamin C synthesis. Only reasonable explanation: the gene was broken long ago in the common ancestor of humans and chimps.

      If you try to hypothesize a creationist explanation, note that this would be a creationist argument, but not an Intelligent Design argument, because the mutations in the pseudogenes have no function and are not designed to do anything. So you can't cry "designed that way", but you could try "created by God that way."

      Then you would have to explain why your God breaks genes and makes them non-functional in the same places and in the same ways in humans and in disgusting animals.

      Delete
    5. Larry says: BTW, the latest RefSeq database has eliminated almost all of the splice variants detected in earlier studies. They have concluded that most of them are artifacts or splicing errors.

      That's a big deal, but do you have a citation on that?

      Delete
    6. Assuming the reference is not Kim et al.

      Delete
    7. Larry quotes Kim et al.: "Alternative splicing gives rise to a large number of splcie variants at the RNA level. Multiple protein isoforms are contributed by only one-third of annotated genes, whereas the remaining two-thirds generate only a single protein product according to the RefSeq database. Although our primary goal was not to obtain complete coverage of splice isoforms, we identified isoform-specific peptides for 2,891 protein isoforms for 2,450 genes."

      In other words, about 12% of human protein-coding genes were shown to produce multiple protein (peptide) isoforms...


      I get how you calculated the 12% (= 2450/20,000), but what is the origin of the discrepancy between 12% and Kim's claim of
      "Multiple protein isoforms are contributed by only one-third of annotated genes"? Is the difference between all genes vs. annotated genes, or because of different criteria for "isoform"?

      Delete
  3. Of course it would be nice if the people claiming that the "tree of life has fallen" would *also* make it clear that is hardly a universally held belief among biologists and that an awful lot of us think a mountain has been made out the lateral transfer molehill. The "humans are amazingly complex and their big genome must mean something" people don't have a monopoly on this sort of distortion.

    ReplyDelete