Talk:Epistasis

Multiple Definitions of epistasis
Epistasis in evolutionary genetics is defined as "departure from multiplicative product" of individual gene's fitness. This is because independent contributions to joined fitness are modelled in multiplicative form. (See publications from Otto and Feldman). The neutral "modifier" and "structural" loci are over-simplified assumptions for the purpose of mathematical modelling.

Epistasis in quantitative genetics is often defined as departure from additive model with respect to a polygenic trait. This is because quantitative trait loci are often modelled in additive form.

Epistasis is largely a genetic term. Its meaning in biology is not well-defined and often leads to confusions. In fact, it is hard to tell which gene is a "modifier" and which one is a "structural" locus, given any real pair of interacting genes. This confusion can be seen in the exchange of arguments in the literature on Hsp90, a candidate "modifier". Hsp90 can reduces phenotypic expression in its substrate genes, satisfying the criteria as a genetical "modifier". Yet, Hsp90 is certainly not a neutral locus. It is an essential gene under strong purifying selection. (See review by Cordell HJ, 2002 Human Molecular Genetics. Epistasis: What it means, what it doesn't mean, and statistical methods to detect it in humans.)

(Postdoc 15:06, 9 September 2007 (UTC))

Incorrect figure
I think that the graph showing epistasis is wrong. I think the y-axis ought to be log fitness (so that in the absence of epistasis each negative mutation causes the fitness to be decremented by a constant factor, rather than a constant amount). This is also how the figure is given elsewhere, e.g. Mark Ridley's 'Evolution', 3rd ed. I don't know how to change the figure (and am writing my thesis for hand-in this week, so don't have the time to find out). Sorry about this - could somebody else change it, please? (CJP)

Figure text:

I changed 'Synergistic' and 'Antagonistic' epistasis in the figure text. As far as I can see it was wrong as it was discribed originally, while I think it was correct in the main text. (Jens Ådne R. Haga)
 * - You are wrong. The slope of the red line decreases as the number of new mutations increases, showing synergistic epistasis. The figure is however confusing, as this line would be expected to be under the black, no epistasis line if all the mutations were of equal effect.
 * I think the figure is correct - red is synergistic - but it does seem inconsistent with the numeric example next to it. That's because the numerical explanation refers to expression of a trait, and the figure refers to fitness. To reconcile the two, you have to realize that the "trait" is a decrease in fitness. So ab shows the wild type fitness, while AB shows the maximal decrease in fitness, with aB and Ab exhibiting fitness values in between (close to WT in synergistic, close to AB in antagonistic epistatis). Confusing! Patrikd (talk) 09:28, 20 February 2008 (UTC)
 * I think the caption is contrary to the figure. The caption says negative epistasis means the mutations are more deleterious in combination than individually, and it says this is shown by the red curve. More deleterious should be illustrated as less fit, which is lower on the Y-axis. However, the red curve is above the black line, indicating higher fitness, at every internal point on the X-axis. This can be fixed by swapping red & blue in the caption or the figure.
 * What matters is not the absolute position of the line, but the change in its slope with the number of mutations. For synergistic epistasis (red), each subsequent mutation is more deleterious than a previous one, i.e., decreases fitness by a bigger amount. For antagonistic epistasis (blue), each subsequent mutation is less deleterious. Yegorka (talk) 09:25, 4 May 2016 (UTC)

Definitions of epistasis
This definition of epistasis seems very limited to me. I thought epistasis means interaction between genes in general. Any ideas? Wilke


 * It's a slippery word: epistasis is sometimes meant in a biochemical sense: two genes (gene products, really) can interact (bind, regulate whatever), or in a statistical genetics/biostatistics sense: i.e. any deviation from genetic additivity. Biochemists, mol. biol. tend to use epistasis in the former sense, and pop. geneticists tend to use the latter definition.   There's an excellent treatment in the chapter "Epistasis" in Lloyd & Fox-Keller's collection Keywords in Evolutionary Biology, Harvard University Press (1992?).


 * I think the article should cover both senses, I have some initial notes that I have been meaning to commit sometime, but I can merge those in later. --Lexor

Moved from article, currently confusingly written and not written in an encyclopedic tone, it's too "chatty", moving here so it can be rewritten: ---Lexor|Talk


 * Classical versus modern genetics


 * At that time there was no idea what a gene could be. One must keep in mind that the structure of DNA was identified in the fifties by Watson and Crick, who won a Nobel prize for this achievement. Nowadays the concept of epistasis may not be of much sense from what Bateson had in mind back in 1909. 


 * As an example, in a simple biosynthesis pathway, if a gene is blocked before the chain end, the steps ahead would probably be shut down. It is somewhat strange to call this as epistasis, but this is the closest idea. So, for molecular genetics, this concept does not help very much, being not that much necessary.


 * The concept of pleiotropy goes on the same line. Actually when one does not understand how a gene can have a broad effect on the phenotype, "just call it as pleiotropic". Regulator genes, like homeoboxes, can have large effects on the phenotype body of the the affected individual: that's because they interfere with the architecture of the body plan. Genes that affect the development of indivials would probably have a "pleiotropic" effect also.

Without consulting the edits, I added a new first paragraph the article demanded (it hardly made sense without it). Some of the previous versions of the page were very consise and accurate. Perhaps rather than re-writing the above excision, we should consider reverting to one of those revisions. --Rritterson

Negative Epistasis
Negative epistasis seems like an unclear concept that is not properly defined. — Preceding unsigned comment added by 128.83.206.217 (talk) 18:55, 14 February 2012 (UTC)

Merge
I think this should be merged with gene interaction too. But maybe epistasis is a more commonly used term, so gene interactions should be merged with this? Unsigned by User:82.35.104.90


 * Yes, that was the conclusion I came to as well. - Samsara contrib talk 00:42, 24 January 2006 (UTC)


 * I changed "gene interaction" to redirect here instead of to "Regulation of gene expression". "Genetic interaction" already redirects here. Dr d12 17:15, 2 January 2007 (UTC)

Section moved to talk page from main article
"Example in Mendelian terms

In a much more detailed example, consider the sweet pea plant. In a simple representation, purple flower color (P) is dominant over white (p). However, consider the addition of a control gene, consisting of two alleles, dominant (C) or recessive (c). In this example, for the flowers to be purple, the plant must have at least one of each dominant allele (i.e. it must be P-C-, where the "-" can be either dominant or recessive).

In a dihybrid cross, such as the pea plant example above, when there is a genetic interaction involved, you often see a modified 9:3:3:1 ratio. Normally, when you have two alleles which assort independently, you get a 9:3:3:1 ratio of phenotypes. Genetic interactions, however, can conceal the ratio, and make it appear to be something other. In the example of the pea plants above, the result is a 9:7 ratio of purple to white phenotypes instead of the expected 12:4 ratio."


 * I moved this section to the talk page because it needs a complete re-write so that the expected ratios can be arrived at by the reader. e.g. What are the phenotypes of CC and cc? What are the parental genotypes expected to produce a 9:3:3:1 ratio? Why would a 12:4 ratio be expected?


 * This could be turned into a good example of using genetic epistasis to order gene products in a biochemical pathway (leading here to pigment) but the article doesn't cover this use of epistasis. Dr d12 02:34, 5 March 2007 (UTC)

For the Layperson reading this, "assort" in the first sentence destroys the meaning of the whole sentence

In the context of medicine, epistasis means something else entirely...
I came across "epistasis" in the context of adverse effects observed in lab animals exposed to certain chemicals. (see for example, page 26 of this document. Obviously, in this context, epistasis doesn't refer to gene interaction. According to dictionary.com, epistasis can also mean "the suppression of a bodily discharge or secretion," which is probably what the term is refering to in this case. It seems that the word can also mean "a film that forms over the surface of a urine specimen." Shouldn't the article refer to these other meanings? Perhaps a disambig is in order.... Yilloslime 17:46, 17 September 2007 (UTC)


 * I briefly looked through the linked paper. To me, "epistasis" in this paper seems to suggest that phenotypes, such as bodily discharge, are suppressed in some strains. This is the classic definition of epistasis, ie, phenotypic variation is masked by genotypic variation at certain loci, which implies gene interaction. Postdoc 17:21, 27 September 2007 (UTC)


 * I disagree. In this document, epitasis is one of the adverse effects associated with poisoning with the chemical being studied. Check out [dictionary.reference.com/browse/epistasis dictionary.com].Yilloslime (t) 15:37, 28 September 2007 (UTC)


 * I see. Interesting definition in Medicine. But, I thought this article implies the concept of Genetics. Postdoc 02:46, 29 September 2007 (UTC)


 * Well that's my point I guess. The article is called Epistasis not Epistasis (Genetics), and as such some folks will invariably come to it looking for info epistasis in the medical context, and be disappointed that this page doesn't provide any info on that. After all, it happened to me. So perhaps there should be a note somewhere in the article saying "In the context of medince...." or perhaps there could even be an Epistasis (Medicine) article, though I'm not the one to write it. Yilloslime (t) 19:11, 29 September 2007 (UTC)


 * I will support such changes. Postdoc 19:08, 1 October 2007 (UTC)

Mistake in the figure showing the difference between antagonistic and synergistic epistasis
In the figure showing the difference between antagonistic and synergistic epistasis the blue line should be antagonistic epistasis and the red should be synergistic epistasis unlike what is stated in the description below the figure.

synergistic epistasis is defined as a situation in which an added deletrious mutation will have a greater effect than the effect the previous mutation had and antagonistic epistasis is defined as a situation in which an added deletrious mutation will have a lesser effect than the previous mutation had. Therefore a curve showing synergistic epistasis (when the Y-axis is fitness and X-axis is the number of mutations) should have an increasing slope whereas the curve of antagonistic epistasis should show a diminishing slope. In the figure shown, the red line that has an increasing slope and therefore should be the depicting synergistic epistasis and the blue line that has a diminishing slope should be depicting antagonistic epistasis. — Preceding unsigned comment added by Nayruru (talk • contribs) 20:33, 4 January 2011 (UTC)
 * My understanding is that the labels in the diagram are correct (the blue bottom curve is synergistic; the red upper curve is antagnostic). Fitness is greatest when no deleterious mutations occur. If some small percentage of such mutations occurs, fitness will decrease, following the black (straight) line if effects are independent (no epistasis). With synergistic epistasis, if some small percentage of deleterious mutations occurs, both the epistatic gene and the modifier genes will be negatively impacted, and the fitness will be reduced more significantly (fitness dropping below the black line). With antagnostic epistasis, the fitness is not reduced as much since whereas the epistatic gene is negatively impacted, so are the modifier genes which will therefore not be so antagnostic. You may be commenting that the slopes of the lines change in the wrong direction. In some big-picture sense the diagram is correct in that when you get a massively large mutation rate, the fitness will be zero for all three curves. Yet I have feeling you might be right about the shape of the curves for small mutation rates (the curves should probably be tangential to the straight line at each end). Johnuniq (talk) 06:41, 5 January 2011 (UTC)
 * I believe the current description is wrong, for the reasons stated above. The figure is similar to that in (Charlesworth and Charlesworth 2010) (ref. 5, cited in the figure caption; p. 549), but the labels are the opposite there: the upward-convex line is synergistic epistasis, and the downward-convex line is antagonistic epistasis. Yegorka (talk) 07:12, 4 March 2014 (UTC)
 * Given the confusion with the graph, perhaps it would be clearer to have the effects of multiple beneficial mutations (i.e. uphill curve)?T. Shafee (Evo&#38;Evo) (talk) 11:05, 8 March 2014 (UTC)

2013 overhaul
I've added quite a bit of content as well as attempting to unify and clarify some of the existing sections, particularly definitions used by different fields. Input would be very much welcomed, especially by a geneticist on any missing terminology and the causes and mechanisms of epistasis between genes. Should hopefully be a better structured article now though. T. Shafee (Evo&#38;Evo) (talk) 08:30, 29 December 2013 (UTC)
 * United classifications to clarify how hey relate to each other
 * Added section on epistasis within genes
 * Added section on causes of epistasis
 * Added section on evolutionary significance
 * Added a stack of references
 * Added images
 * Thanks to CharlesInVermont for input on the genetics side of things (particularly diploids). I've reformatted a bit but tried to keep the information.T. Shafee (Evo&#38;Evo) (talk) 08:36, 12 January 2014 (UTC)
 * With all the updates and improvements I'd like to upgrade the article from Start class to B class. Hopefully others agree.T. Shafee (Evo&#38;Evo) (talk) 04:26, 14 January 2014 (UTC)

The baldness and hair color example of epistasis
This illustration should be removed from the article because it is incorrect. The article on Baldness says that those genetically predisposed to baldness lose hair on top of the head, leaving a bald top and hair growing around the head below it. Obviously someone with male pattern baldness could still have hair of one color or another. No "baldness gene" is linked from this illustration which causes complete absence of body hair. Does such a gene exists, so that the result is like the effects of radiation or chemotherapy? If so, identify it and link to it. Otherwise the dubious illustration should be removed from the article. (Maybe there is a gene which makes a guy or gal shave their head so they look like Kojak). Edison (talk) 20:21, 26 May 2014 (UTC)
 * Thanks for the comment, and sorry for the slow response! The baldness I was thinking of when I included the image was Alopecia totalis or Alopecia universalis. I used baldness as an easy-to-understand trait so that people just reading the header weren't confronted with mutations in biochemical pathways or enzymes. I'll include the link to Alopecia totalis (since it has the better article) in the bald figure legend. I'm also happy to draw up other suggestions of simple-to-understand epistasis examples! T. Shafee (Evo&#38;Evo) (talk) 12:18, 12 December 2014 (UTC)

Positive/negative epistasis for beneficial/deleterious mutations
I've done another check of the literature to correct the definitions of positive and negative episatsis. The classifivations section and the figure legend of the line graph were incorrect for deleterious mutations The previous definitions were as follows:
 * 1) With beneficial mutations - positive epistasis is a greater-than-additive effect of two beneficial mutations.
 * 2) With deleterious mutations - positive epistasis is a greater-than-additive deleterious effects

However statement (2) is incorrect (see quotes below) "Where deleterious mutations are concerned, positive epistasis lessens the fitness reduction predicted from individual mutational effects" Prevalent positive epistasis in Escherichia coli and Saccharomyces cerevisiae metabolic networks - Nature Genetics 42, 272–276 (2010)

"negative epistasis, whereby the combined effects of mutations are larger than expected from their individual effects" Stability effects of mutations and protein evolvability - Current Opinion in Structural Biology 19, Issue 5, (2009)

I think that statement (1) is still correct though:

"Such interactions are referred to as positive epistasis, in which fitness of a multiple-mutant genome is higher than those of the individual mutants" ''Viral quasispecies evolution - Microbiol. Mol. Biol. Rev. 76, 159–216 (2012).''

"The interactions within the positive epistasis class, which provide greater than expected resistance to MMS, were used..." Epistasis - the essential role of gene interactions in the structure and evolution of genetic systems - Nature Reviews Genetics 9, 855-867 (November 2008)

I've updated the article, line chart figure legend, and whole barchart figure to reflect this. To confirm - positive epistasis amongst deleterious mutations gives smaller fitness drops than expected.

T. Shafee (Evo&#38;Evo) (talk) 11:57, 12 December 2014 (UTC)


 * I've now further revised the text, clarifying positive/negative vs. synergistic/antagonistic epistasis, and making the figure caption consistent with the main text. The preceding confusion arose because these are in fact two different classifications. Yegorka (talk) 17:48, 13 November 2015 (UTC)