A genomic predictor of lifespan in vertebrates

https://www.nature.com/articles/s41598-019-54447-w

by Vaslo • 7 years ago

117 77 7 years ago

nature.com

tpmx• 7 years ago

That article title is crap.

Edit: the title is fixed now.

alexgmcm• 7 years ago

I agree, that title is awful given the actual link.

mullingitover2• 7 years ago

In the past 200 years, the average life expectancy of humans has more than doubled because of modern medicine and changes in lifestyle

I thought this idea that humans had short lifespans historically was thoroughly debunked. Humans have historically had high infant mortality rates which threw the numbers off, but if you survived past the age of three you were pretty likely to make it to your sixties or seventies.

pharrington• 7 years ago

Your second sentence directly contradicts your first.

mullingitover2• 7 years ago

Maybe a better way of putting it is that averages are a garbage statistic when looking at these numbers, because they hide the fact that the normal human lifespan is and always has been roughly the same.

psychometry• 7 years ago

Mortality rate actually reaches its minimum around age 10 or 11, so you would see even longer lifespans if you only look at people who survive at least to their teens.

3fe9a03ccd14ca5• 7 years ago

Our days may come to seventy years, or eighty, if our strength endures; yet the best of them are but trouble and sorrow, for they quickly pass, and we fly away.

How old is this proverb?

hprotagonist2• 7 years ago

youngest is within a 500 year window around the date of the composition of the Psalms (it's psalm 90)

https://en.wikipedia.org/wiki/Psalms#Origins

philipkglass4• 7 years ago

It's true that infant mortality has decreased faster than other mortality in the past century. Even people who survived past infancy had significantly higher mortality rates at the beginning of the 20th century, though.

See "Table 6 - Period Life Tables for the Social Security Area by Calendar Year and Sex" for detailed information from the United States:

https://www.ssa.gov/oact/NOTES/pdf_studies/study120.pdf

In the table series for 1900, male life expectancy at birth was 46.4 years and female life expectancy was 49 years. For the survivors of infant mortality, additional life expectancy at age 6 was 53.2 years and 54.3 years respectively. Most children could not expect to reach their 60th birthday even if they had survived infancy.

At age 18, remaining life expectancy was 41.7 and 42.9 years, for total expected age at death of 59.7 and 60.9 years respectively; weighted by sex distribution at age 18, these young adults had an average life expectancy of 60.3 years.

ColanR3• 7 years ago

I don't think that looking at lifespans for the industrial revolution has any bearing on what lifespans used to be 'historically'. It was pretty clear to me, from reading the GP that they were talking about pre-industrial revolution times.

melling• 7 years ago

Aren’t we talking about two different things?

“Maximum lifespan is believed to be under genetic control”

Given modern medicine etc. more people are approaching the maximum lifespan, but we aren’t increasing the maximum.

Some people in the middle ages lived well into their 80s, just not nearly as many as today.

For example, 600 years ago Michelangelo lived to be 88. https://en.m.wikipedia.org/wiki/Michelangelo

abj• 7 years ago

You're right it's debunked, but even if you take survival rate at age 15 in the Paleolithic era life expectancy was 54 [1]. There were some aristocratic populations that last over 64 at age 21 [1].

[1] https://en.m.wikipedia.org/wiki/Life_expectancy

caiocaiocaio• 7 years ago

I've certainly seen that on internet, but I don't known of any real research or hard data, and just go to any old graveyard and work out people's ages. Seems like clickbait debunking rather than real debunking.

duelingjello• 7 years ago

I'd say that lots of kids and infants dying in the past dragged down the average to seem comparatively low compared to today. Of those who lived past 18, I'd wager most adults lived almost as long as today unless an infection or simple mechanical injury happened.

Cancer, Alzheimer's, CHD, diabetes' complications and pneumonia gets most people these days. If you live long enough these days, either your heart will give out, your circulatory system will leak and you'll die from shock, you'll get pneumonia or an infection. Finally, the vast majority of end-of-life operations are worse than nothing at all by worsening quality-of-life... hospice is better in most circumstances.

PS: follow the holistic/comprehensive list of preventative lifestyle changes for Alzheimer's so you can go out with your mind as intact as possible and stay fit/able-bodied to where you don't need long-term care (LTC). As an example, the grandpa of that young guy from that show Gold Rush; what a full- and active-life that guy had.

wetpaws• 7 years ago

Life expectancy != maximum lifespan

otabdeveloper2• 7 years ago

I thought this idea that humans had short lifespans historically was thoroughly debunked.

Then the Lord said, “My Spirit will not contend with humans forever, for they are mortal; their days will be a hundred and twenty years.”

Genesis 6:3.

I don't know how old Genesis really is, but 2000 B.C. might be a good enough guess. So yeah, pretty thoroughly.

shele• 7 years ago

If the “natural lifespan” of humans ends to be 38 you are using a flawed definition of natural. This also connects to people believe that humans in the past died with 30 latest because the life expectancy was 26 years, ignoring that this is an average heavily influenced by child mortality

raspasov2• 7 years ago

Can we please repeal and replace the title of this with “A genomic predictor of lifespan in vertebrates” (aka the real title) ?

dang• 7 years ago

Yes. The submitted title was "Humans have a natural lifespan of only 38 years", which broke the site guidelines. They ask: "Please use the original title, unless it is misleading or linkbait; don't editorialize." We've reverted the title now.

https://news.ycombinator.com/newsguidelines.html

raspasov2• 7 years ago

thumbs up

krustyburger• 7 years ago

Early humans have been reported to have a maximum life expectancy of 40 years, less than half by modern standards. Similarly, in chimpanzees the lifespan was estimated at 39.7 years. The maximum longevity of a chimpanzee in the wild is thought to be of a 55 years old female, however it is reported that many live to approximately 40 years of age.

It seems like what we’re really seeing here is the similarity in lifespan between our earliest human ancestors and other primate species.

The earliest humans would not yet have had access to consistent shelter, fire or tools/weapons and these developments carried with them enormous benefits with respect to health and comfort.

hprotagonist2• 7 years ago

previously https://news.ycombinator.com/item?id=21773562

ksydbd_383838• 7 years ago

At a high level, I am very skeptical. Usually when your model does not agree with observations, the problem is with the model. A few things that stand out to me. I have not thought deeply about any of them, so please correct me if I am mistaken about any of these.

This primary data set contained 252 species from five vertebrate classes [...] We removed humans (Homo sapiens) from the data set as they were listed with a maximum lifespan of 120 years, which does not reflect the variability and the true global average lifespan (60.9–86.3 years)

So why should we trust the rest of this dataset? Garbage in garbage (GIGO) out comes to mind.

We used promoter sequences centred around the transcription start site (TSS) (-499 to 100 bp of each promoter) in Humans (Homo sapiens) from the EPD as the data set of promoter sequences. [...] Briefly, as described previously, using Basic Local Alignment Search Tool (BLAST) v2.2.31 the promoter sequences were mapped to the single top hit in each species.

This would seem to imply a weird correlation structure between data examples that could pose problems for training/test split and/or linear models. I would also liked to see some QC where they show how well this recovered known (i.e. annotated) promoter regions. Are they picking up false positives? Are they missing stuff?

The glmnet function was set to a 10-fold cross validation which returns the best performing model. [...] This resulted in a total of 42 promoters for estimate lifespan.

So they're doing post-selection inference, so p-values are suspect. Tibshirani (inventor of Lasso) and Taylor recently released a package for post-selection inference, which I do not see them using here.

Species were randomly assigned to either a training (176 samples) or testing (76 samples) data set (70/30 split).

Rule of thumb: you usually want about number of example = 10x number of features to avoid overfitting. 42 features seems kind of thin. Even worse when you consider that there might be a correlation between training and test examples imposed by the initial selection of promoter sites using BLAST.

ramblenode• 7 years ago

> We used promoter sequences centred around the transcription start site (TSS) (-499 to 100 bp of each promoter) in Humans (Homo sapiens) from the EPD as the data set of promoter sequences. [...] Briefly, as described previously, using Basic Local Alignment Search Tool (BLAST) v2.2.31 the promoter sequences were mapped to the single top hit in each species.

Measurement error should probably be modeled.

derefr3• 7 years ago

So, is this suggesting that humans, as an animal species, are basically "engineered" by evolution to have a 38-year flat part of their MTBF bathtub-curve?

Where I mean "engineered" in the sense that evolution only can only really steer toward adaptive fitness of an organism up until a certain point in its lifetime, that being the point when the organism has done all the breeding they're going to do. Past that point, most things the organism does won't impact the evolutionary cohort of its species, so, turning that around, there will be no adaptations in its genome to help it survive past that point (since, where would they be sourced from?)

That time coming at 38 years for humans would explain a lot, I think. (It makes sense; that's about the oldest average age that humans are willing to continue to have children at.)

HarryHirsch• 7 years ago

when the organism has done all the breeding they're going to do

You could argue that having older and more experienced individuals around might increase chances for survival for everyone. You'd predict that that effect would be especially pronounced in a species that highly relies on culture, like us humans. Indeed the hypothesis was found true for orcas: https://www.pnas.org/content/early/2019/12/03/1903844116

derefr3• 7 years ago

It can be true, but for a gene to reach fixation, it has to be selected for, i.e. organisms without it have to die before breeding.

So this effect might work out if you have a lot of isolated sub-populations, and the ones that live less long suffer extinction, while the ones that live longer go on to eventually reconnect and breed. But it doesn’t really work in a “liquid market”, because free riders without the longevity genes will be just as successful at breeding (due to the existence of the long-lived species members) as the ones with those genes.

landryraccoon• 7 years ago

Wouldn't evolution steer animals even after breeding if the parents continue to contribute to the success of their children?

If having long lived parents was beneficial to their offspring, evolution will select for longevity well past breeding age. OTOH, if an organism laid eggs and then vanished from it's offspring's lives forever, then it would be hard to see how evolution would select for longevity past the egg laying phase.

JoeAltmaier• 7 years ago

Emphatically yes! I believe that may even be the reason humans grow feeble when old - so that the grandparents sit around the fire and nurture the young, teaching and telling stories.

downrightmike• 7 years ago

That's why we see menopause in us and in other species like Orcas. Grand parents contribute more food to their grand children without competing to have more kids. I can't find the reference, Google is spammed with diet websites.

wahern• 7 years ago

You're only 1/4 related to grand children. A strategy of living a long life to help raise a grandchild means that 3/4 of your effort is promoting someone else's genes. Now what if the strategy of those other people is to have many more children, possibly with many different partners, with a relatively diminished per-child investment, reliant on someone like you to help raise those children. Who's going to out-compete whom?

I'm not saying the second strategy is better, just pointing out that the first strategy is self-limiting. Which means we can't even begin to understand the benefits and viability without answering more complex questions; questions we don't yet have answers to.

Also, humans are the only known species, extant or extinct, which exhibits significant non-kin altruism.[1] There's no strong theory for how this emerged. Which means there are some very important dynamics to human evolution (and evolution in general) that we're completely ignorant to--we don't even know what the questions are, let alone the answers.

[1] The most popular mammals used for comparison to humans, bonobos and naked mole rates, are organized as matriarchies--the females are sisters, somewhat like ants and bees. The above-average altruism they exhibit is easily explained by basic Darwinian genetic evolutionary theory.

allovernow4• 7 years ago

You know, a man who was young indefinitely would have a lot of time to spread his seed...I'd expect such a trait to be strongly selected for if it arose. Unless it had a stiff penalty, maybe more competitive among men, or interbreeding problems?

derefr3• 7 years ago

Right. Think of “breeding” as the complete act of making autonomous organisms. If your children can’t survive without you, then in the evo-bio sense, you haven’t finished “breeding” yet, because the ability of your genes to be passed on still depends on facts about your genome rather than solely on facts about your child’s genome.

Humans take a pretty long time to “breed” compared to most animals, since humans are born pretty non-autonomous compared to most animals. But there still comes a point where the probability of breeding success of a child becomes entirely dominated by the factor of that child’s genes, over the factor of the parent’s.

Also note that, just because this effect exists, doesn’t mean it guarantees that it will be an effect in the direction of longer health-span in every organism. In fact, in many organisms, the organism is wired to die as part of the breeding process, since dying better guarantees success for the offspring! (Some because they no longer compete with their offspring for food; some because their body decays and fertilizes the spawning grounds, leading to a bountiful harvest for their children; etc.)

valw• 7 years ago

To the experienced data scientists / statisticians here: does this sort of regression analysis between the actual and predicted age seem legit to you?

It seems very weird to me. From what I understand, the p-value tests the hypothesis that the predicted and actual data are correlated, which seems like a very weak way of assessing the reliability of the prediction.

An estimator of the test error seems more relevant, and that's kind of what the R^2 does, but why do an affine regression between predicted and actual value instead of say an RMS error? Isn't this using the test data for parameter-fitting, i.e training?

ColanR3• 7 years ago

Off topic, but based on the discussions here.

I'm seeing a lot of discussion here about how the lifespan of humans (not counting infant and teen mortality rates) may not have improved much in the last thousand years. As a tangential question, does that mean all our medical progress has a) improved infant mortality and b) counteracted the unhealthiness of modern life, and not done much for us beyond that?

lordnacho• 7 years ago

From what I've read that's right. Kids don't die on nearly the same scale, likewise mothers giving birth.

Nutrition has gotten better, which helps immune system strength.

We know what causes a lot of diseases and can protect against their spread. Both reactively in terms of illnesses people get and in terms of prevention such as vaccines and warnings about smoking.

Even then there are records of famous ancient people like Egyptian kings who lived to what we'd still consider a ripe old age now.

iten• 7 years ago

I recently completed my PhD in vertebrate comparative genomics so this is fun to see.

The single most important factor that needs to be accounted for in analyses like these is the correlation between phylogenetic similarity and the trait in question. In short, closely related species will tend to have similar lifespans, and closely related species will tend to have similar CpG density in any fixed genomic region. So the fact that you can predict lifespan from CpG density with enough parameters is unsurprising. You could almost certainly predict lifespan fairly well from any feature measuring phylogenetic similarity -- I would have liked to see some evidence showing that CpG density in these promoters is somehow uniquely suited for the task.

rflrob• 7 years ago

I wonder whether they also looked at body size. Larger species tend to live longer, so it could just be that their regions are just associated with growth.

asdff3• 7 years ago

They did not.

asdff3• 7 years ago

They did mention they were looking at a couple of long lived examples like rockfish that live 200 years vs killifish that live for 1, so this seems to be variable among fish at least. As for mechanism, they were pretty hand wavey but cited reference 33 and mentioned that longer CpG tracts are thought to be protective against spontaneous methylation. Unmethylated CpG islands in promotors are associate with active transcription, especially found in housekeeping genes that are transcribed constantly by definition. Methylating these islands can silence the gene. Makes sense but needs an experiment to know for sure.

jmpman• 7 years ago

No comments about dog lifespans? Come on genetic engineering, give me a Labrador that lives 20 years and doesn’t shed.

asdff3• 7 years ago

Get a hairless cat, similar personality actually

A genomic predictor of lifespan in vertebrates

Related Stories