Continuing the discussion from Theman8469’s Complaints about Nathan Lents:
The original quote from Sternberg…
The segments of our DNA that are commonly called “genes” consist of protein-coding exons and non-protein-coding introns. Initially, the entire DNA segment is transcribed into RNA, but between ninety and ninety-five percent of the initial RNAs are “alternatively spliced.”
What is alternative splicing? Imagine that the initial RNA derived from its DNA template has the organization A—B—C—D—E—F, where the letters represent blocks that specify amino acid sequences and the dashes in between the letters stand for introns. Alternative splicing enables multiple proteins to be constructed given the same RNA precursor, say, ABCDF, ACDEF, BCDEF, and so forth. In this way, hundreds or thousands of proteins can be derived from a single gene.
There’s more. The messenger RNAs that are produced by this process—and therefore the proteins that are made in a cell—are generated in a way that depends on the stage of development as well as the cell and tissue type. In the above example, a nerve cell may express the ACDEF version of a messenger RNA whereas a pancreatic cell may produce only the BCDE version. The differences are biologically essential.
What does this have to with introns? Everything. It is the presence of introns that makes this permutative expansion of messenger RNAs possible in the first place.
So let’s do the math. At least ninety percent of gene transcripts undergo alternative splicing, and there are at least 190,000 introns in the human genome. That means we have at least 0.90 x 190,000 = 171,000 introns that participate in the alternative-splicing pathway(s) available to a cell.
Larry Moran responds:
It’s up to you, dear readers, to figure out all the things wrong with this explanation. You can start with the math. Arithmetic isn’t one of their strong points. Or maybe it’s an understanding of biology that’s the real weak point?
@theman8469 rightly protests that 0.90 x 190,000 really does equal 171,000. Larry knows this too. That is not what biologists mean by “math errors”, though I fully admit this is our idiosyncratic way of talking. We often say that some one is making a “math error” when they are computing the wrong quantity with the wrong assumptions at play. This is not really an “arithmetic” error, even though we often say it is. Rather, it is an error in mathematical modeling.
This is how Larry explains it when pressed on this…
There are two important problems with Richard Sternberg’s article.
First, it’s almost certainly not true that 90-95% of all human protein encoding genes exhibit alternative splicing. While there’s no consensus right now, the majority of biochemists think this number is too high. Whether it’s 5% (my guess) or 50% (a common estimate) is controversial.
Sternberg gives us no indication that he understands this controversy. He bases his entire fairy tale on a value that has been pretty much discredited.
His second problem is not even understanding the consequences of his false assumption. Let’s assume that Sternberg is correct and 90-95% of all protein encoding genes exhibit alternative splicing. Since there are 20,500 such genes in our genome it follows that about 19,000 are alternatively spliced.
The minimum requirement for alternative splicing is that an intron can be included or excluded in the mRNA. The lower limit for the number of introns is thus 19,000 (actually 18,450 - 19,475).
In other words, as many as 171,000 introns could be (mostly) useless junk. It will be even more if we accept a lower number of alternatively spliced genes.
I don’t think this is the message that Sternberg meant to convey in his posting. This is why we call them IDiots.
BTW, Sternberg took the 190,000 introns information from Stephen Matheson who quotes a 2005 paper by Fedorva and Fedorov. According to those authors, there are 23,506 protein encoding genes in our genome of which 21,746 contain introns. Thus, the average number of introns per gene is greater than 8.
We now know there are fewer genes and the total number of introns has been dropping steadily as workers begin to eliminate spurious introns when genes are annotated. I think there are about 150,000 introns (see: Junk in Your Genome: Protein-Encoding Genes).
The exact number doesn’t matter so much but the point is you can’t mix and match numbers. It’s invalid to use a lower estimate of introns per gene (e.g. 7.6) with a high estimate of the number of introns (e.g. 190,000).
Notice, at no point does Moran question that 190 x 0.9 = 171. Rather, he is questioning the premise of the calculation, saying the wrong numbers are being multiplied together and that Sternberg is making a poor estimate.
I entirely grant that this is not, literally, an “arithmetic” error in the most strict of senses. Though, plugging the wrong numbers into a formula is sometimes called on an “arithmetic” error. Errors in mathematical reasoning or modeling, like this, are often called “math errors” by biologist. That is our jargon. Yes, it can lead to confusion. I wonder though if similar jargon is found among @physicists.