Epigenetic Inheritance in Humans

[Thermos aquaticus]

I recently came upon a website (via Jerry Coyne's blog) that reviews some of the most cited peer reviewed papers touting evidence for transgenerational epigenetic inheritance in humans. As the reviewer, Kevin Mitchell, discusses, there really isn't any evidence that epigenetics is inherited across generations in humans:

Grandma’s trauma – a critical appraisal of the evidence for transgenerational epigenetic inheritance in humans

The common problem with these studies is that they lack a priori hypotheses which is compounded by small population sizes. What does this mean? It comes down to the old chestnut, "correlation does not necessarily imply causation". Whenever you have a ton of data for many different characteristics it is almost guaranteed you will find a positive correlation between two of those characteristics even when there is no cause that links them. This is especially true for small sample sizes, which is the case in these epigenetic studies. What they have done is throw the data at the wall to see what sticks.

It is proper to do these correlations at the start of a research project. The second step would be to find a cause for the correlation, if there is one. However, no cause is given in these papers, just the correlation. That's a problem. This isn't to say that transgenerational epigenetic inheritance doesn't happen in humans, but there certainly isn't any strong scientific evidence for it at this time.

So the real question is why is human epigenetic inheritance such a hot topic in many circles if there isn't strong evidence to support it?

 

[John53]

Is there a simple lay persons description of epigenetic inheritance? I tried googling it and read a few different pages but I still have no idea what it is.

 

[Thermos aquaticus]

Is there a simple lay persons description of epigenetic inheritance? I tried googling it and read a few different pages but I still have no idea what it is.

The most common epigenetic mechanism is DNA methylation. This is where methyl groups are added to the DNA bases in an "island" of DNA sequence that has a lot of CG's in it (called CpG's). So something like ACGCGTCGTTCGCGA. If these CpG islands are methylated upstream of a gene it can reduce the expression of that gene. Since DNA methylation does not involve a change in the actual DNA sequence it is considered to be epi-genetic, as in above-genetics. DNA methylation can also be caused by environmental cues, so DNA methylation can be influenced by the environment.

Some have hypothesized that DNA methylation patterns can be passed on to subsequent generations, meaning that the environment can cause changes in a species. This runs counter to the standard view of evolution which proposes mutational changes that are not influenced by the environment.

 

[ADigitalArtist]

Is there a simple lay persons description of epigenetic inheritance? I tried googling it and read a few different pages but I still have no idea what it is.

To add to that, epigenetic inheritance is of great interest to people in medicine because it can help explain why inheritable disease is not just confined to 'dominant' and 'recessive' but can be both factors of genetics and factors of environment at the same time.

 

[Thermos aquaticus]

To add to that, epigenetic inheritance is of great interest to people in medicine because it can help explain why inheritable disease is not just confined to 'dominant' and 'recessive' but can be both factors of genetics and factors of environment at the same time.

In such cases the epigenetic factor is secondary. The primary cause would still be inherited DNA sequence since the DNA methylation patterns are not inherited.

To use an analogy, it's a bit like having a weak spot in a balloon. The balloon pops when you put air into it because it started out with a weak spot. In this analogy the methylation is the air and the weak spot is the inherited DNA mutation.

 

[ScottySatan]

Having spent 12 years doing biomedical science and writing papers, I think I have something to offer.

The common problem with these studies is that they lack a priori hypotheses

The common problem is exactly the opposite: that these studies do have a prior hypothesis. Results must support the hypothesis. If they don't do that, then the study won't get published. Journals don't publish "negative results". This creates a powerful incentive for bias. That the researchers had a desire that the data support their hypothesis gets downplayed in a publication.

I can't say that this is what happened in this specific paper, but it's almost always true.

It is proper to do these correlations at the start of a research project. The second step would be to find a cause for the correlation, if there is one. However, no cause is given in these papers, just the correlation. That's a problem.

Proof. That's the rub (I know that's not the word you used, I brought it in myself). It's almost impossible to prove that something is true. This usually comes from a massive body of literature, hundreds of papers, not ONE paper. Not in biology.

The closest that biology can come to showing causation is to say that "A is necessary and sufficient for B". That happens in a single paper very rarely, I can't think of an example of any paper that does all of this. Because of the "publish or perish" thing in academia, you must publish more frequently, not waiting until the conclusion of the story. The one paper with the incomplete story such as this will have taken probably at least 18 months of work.


Epigenetics being a hot topic? Meh. Papers on genetics proper outnumber epigenetics papers at least 1000:1. Epigenetics is getting more popular than before because it's consistently being found to have a bigger impact than previously thought. That's pretty scientifically interesting.

 

[Guy Threepwood]

Having spent 12 years doing biomedical science and writing papers, I think I have something to offer.



The common problem is exactly the opposite: that these studies do have a prior hypothesis. Results must support the hypothesis. If they don't do that, then the study won't get published. Journals don't publish "negative results". This creates a powerful incentive for bias. That the researchers had a desire that the data support their hypothesis gets downplayed in a publication.

I can't say that this is what happened in this specific paper, but it's almost always true.



Proof. That's the rub (I know that's not the word you used, I brought it in myself). It's almost impossible to prove that something is true. This usually comes from a massive body of literature, hundreds of papers, not ONE paper. Not in biology.

The closest that biology can come to showing causation is to say that "A is necessary and sufficient for B". That happens in a single paper very rarely, I can't think of an example of any paper that does all of this. Because of the "publish or perish" thing in academia, you must publish more frequently, not waiting until the conclusion of the story. The one paper with the incomplete story such as this will have taken probably at least 18 months of work.


Epigenetics being a hot topic? Meh. Papers on genetics proper outnumber epigenetics papers at least 1000:1. Epigenetics is getting more popular than before because it's consistently being found to have a bigger impact than previously thought. That's pretty scientifically interesting.

Yes, we have at least established that 'random mutation' of the gene sequence is not the exclusive omnipotent driver of change as once believed in some circles, that's progress. The simplest explanation is ever the most tempting, but nature has shown little regard for Occam's razor.

 

[Thermos aquaticus]

The common problem is exactly the opposite: that these studies do have a prior hypothesis. Results must support the hypothesis. If they don't do that, then the study won't get published. Journals don't publish "negative results". This creates a powerful incentive for bias. That the researchers had a desire that the data support their hypothesis gets downplayed in a publication.

That's not what I am seeing. They were simply looking for correlations with no specific correlations in mind. For example, they didn't hypothesize that grandparents who experience severe starvation and or trauma would have grandchildren who were smaller, taller, increased rates of heart disease, decreased rates of heart disease, or any other parameter. Instead, they collected the data and then pumped it through statistical software to see what correlations there were.

Proof. That's the rub (I know that's not the word you used, I brought it in myself). It's almost impossible to prove that something is true. This usually comes from a massive body of literature, hundreds of papers, not ONE paper. Not in biology.

The closest that biology can come to showing causation is to say that "A is necessary and sufficient for B". That happens in a single paper very rarely, I can't think of an example of any paper that does all of this. Because of the "publish or perish" thing in academia, you must publish more frequently, not waiting until the conclusion of the story. The one paper with the incomplete story such as this will have taken probably at least 18 months of work.


Epigenetics being a hot topic? Meh. Papers on genetics proper outnumber epigenetics papers at least 1000:1. Epigenetics is getting more popular than before because it's consistently being found to have a bigger impact than previously thought. That's pretty scientifically interesting.

I am fine with using "proof" with an implied "beyond a reasonable doubt". That is the way most people use the term, so I don't see a problem as long as it is clear that the "beyond a reasonable doubt" is a part of it.

I also don't think anyone is looking for a massive giant piece of proof for epigenetics. What they are looking for is condition X stimulates gene Y which methylates promoter Z and results in phenotype A. Just one pathway would be of interest, but I am not seeing much of that in these papers.

I also think that epigenetics has captured the imagination of the general public which is why it is probably more popular outside of science than it is inside of science. Epigenetics woo is starting to eclipse quantum woo.

 

[Thermos aquaticus]

Yes, we have at least established that 'random mutation' of the gene sequence is not the exclusive omnipotent driver of change as once believed in some circles, that's progress. The simplest explanation is ever the most tempting, but nature has shown little regard for Occam's razor.

Random mutations that change DNA sequences are the dominant driver of inherited change. That's the point that is being made.

 

[Guy Threepwood]

Random mutations that change DNA sequences are the dominant driver of inherited change. That's the point that is being made.

what do you feel is the most compelling example of change being directly witnessed (empirically observed)?

 

[Thermos aquaticus]

what do you feel is the most compelling example of change being directly witnessed (empirically observed)?

There are studies where they sequence the genomes of parents and their offspring to find new mutations in offspring.

Mutation of the DNA molecule is one of the most fundamental processes in biology. In this study, we use 283 parent-offspring trios to estimate the rate of mutation for both single nucleotide variants (SNVs) and short length variants (indels) in humans and examine the mutation process. We found 17812 SNVs, corresponding to a mutation rate of 1.29 × 10−8 per position per generation (PPPG) and 1282 indels corresponding to a rate of 9.29 × 10−10 PPPG. We estimate that around 3% of human de novo SNVs are part of a multi-nucleotide mutation (MNM), with 558 (3.1%) of mutations positioned less than 20kb from another mutation in the same individual (median distance of 525bp). The rate of de novo mutations is greater in late replicating regions (p = 8.29 × 10−19) and nearer recombination events (p = 0.0038) than elsewhere in the genome.
Multi-nucleotide de novo Mutations in Humans

 

[Guy Threepwood]

There are studies where they sequence the genomes of parents and their offspring to find new mutations in offspring.

Mutation of the DNA molecule is one of the most fundamental processes in biology. In this study, we use 283 parent-offspring trios to estimate the rate of mutation for both single nucleotide variants (SNVs) and short length variants (indels) in humans and examine the mutation process. We found 17812 SNVs, corresponding to a mutation rate of 1.29 × 10−8 per position per generation (PPPG) and 1282 indels corresponding to a rate of 9.29 × 10−10 PPPG. We estimate that around 3% of human de novo SNVs are part of a multi-nucleotide mutation (MNM), with 558 (3.1%) of mutations positioned less than 20kb from another mutation in the same individual (median distance of 525bp). The rate of de novo mutations is greater in late replicating regions (p = 8.29 × 10−19) and nearer recombination events (p = 0.0038) than elsewhere in the genome.
Multi-nucleotide de novo Mutations in Humans

which corresponded to what? the kid having a slightly bigger nose than his dad?

To be clear, I'm asking what you think is the best example of observed consequential, morphological , advantageous change actually occurring through 'random mutation', being observed directly by humans

 

[Thermos aquaticus]

which corresponded to what?

The direct observation of mutations occurring in the human genome.

To be clear, I'm asking what you think is the best example of observed consequential, morphological , advantageous change actually occurring through 'random mutation', being observed directly by humans

Those would be the mutations which separate us from chimps.

 

[Guy Threepwood]

The direct observation of mutations occurring in the human genome.



Those would be the mutations which separate us from chimps.

a species of chimp has been directly observed evolving into a human?

 

[Thermos aquaticus]

a species of chimp has been directly observed evolving into a human?

We directly observe that the genetic differences between chimps and humans are responsible for the physical differences between those species. We also directly observe that the mutations occurring human genomes match the pattern of mutations that separate chimps and humans, so we have very strong evidence from direct observations that the differences between chimps and humans are due to the observed process of mutation.

 

[Guy Threepwood]

We directly observe that the genetic differences between chimps and humans are responsible for the physical differences between those species. We also directly observe that the mutations occurring human genomes match the pattern of mutations that separate chimps and humans, so we have very strong evidence from direct observations that the differences between chimps and humans are due to the observed process of mutation.

So that's a no...

I understand the theory, the hypothetical extrapolation, but again, what do you think is the best example of morphological change actually occurring and being directly observed scientifically?

anything?

 

[Thermos aquaticus]

So that's a no...

How are we not observing random mutations occurring right now in human populations?

I understand the theory, the hypothetical extrapolation, but again, what do you think is the best example of morphological change actually occurring and being directly observed scientifically?

Why do you think chimps and humans are different from each other? Is it because of the differences in DNA between their genomes?

 

[Guy Threepwood]

How are we not observing random mutations occurring right now in human populations?



Why do you think chimps and humans are different from each other? Is it because of the differences in DNA between their genomes?

Sure, just as two internet forums are different because of their code, I'm skeptical that one morphed into the other by accidental copying errors, in part because it has never been observed to happen.


So nothing at all? no good directly observable instances of an advantageous morphological change actually occurring that you can think of?

 

[Thermos aquaticus]

Sure, just as two internet forums are different because of their code, I'm skeptical that one morphed into the other by accidental copying errors, in part because it has never been observed to happen.

So what are the differences between the chimp and human genomes that you think random mutations could not produce?

 

[Guy Threepwood]

So what are the differences between the chimp and human genomes that you think random mutations could not produce?

The most difficult improvements to account for by chance are mental capacities, doubling brain size in an apparently short space of time, acquiring innate capacities for language, awareness, abstract thought-- all through blind luck of random copying errors?

losing attributes- like hair, strength, or eyes in cave fish, wings on flightless birds, that's the sort of change that is more easily explained by unguided random corruption of information- but identifying verifiable cases of morphological improvements being introduced, is very difficult as we have established here.