Thursday, February 26, 2015

Digesting yeast's message

A new paper in Nature by Levy et al. reports on the genomic consequences of large-scale selection experiments in yeast.  Yeast reproduce asexually and clones can be labeled with DNA 'barcode' tags and followed in terms of their relative frequency in a colony over time.  This study was able to deal with very large numbers of yeast cells and because they used barcodes the investigators could practicably follow individual clones without needing to do large-scale genome sequencing.  Prior to this, this sort of experiment was prohibitively costly and laborious.  So the authors add to findings in selection experiments using bacteria or flies and so on, where mostly aggregate responses could be identified.

In this case, nutrient stress was imposed, and as beneficial mutations occurred and gave their descendant cells (identified by their barcode) an advantage, the dynamics of adaptation could be followed.  The authors showed, in essence, that at the beginning the fitness of the overall colony increased as some clones, bearing advantageous mutations, rose rapidly in relative frequency.  Then, the overall colony fitness stabilized and subsequent advantageous mutations were largely kept at low frequency (most eventually went extinct).  But overall, the authors found thousands of colonies with different advantageous variants; most fitness effects were of only a small (or, for the majority, very small) percent.  Once a set of large numbers of 'fit' variants had become established, new ones had a difficult time making any difference, and hence staying around very long.

This study will be of value to those interested in evolutionary dynamics, though I think the interpretation may be rather more limited than it should, for reasons I'll suggest below.  But I would like to comment on the implications beyond this study itself.

Who cares about yeast (except bakers, brewers, and a few labs)?  You should!
This is interesting (or not) you might say, depending on whether you're running a yeast lab, or in the microbrew or bakery business. But there are important lessons for other areas of science, especially genomics and the promises being made these days.  Of course, the lesson isn't a pleasant one (which, you might correctly assume, is why we're writing about it!).

This study has important implications for basic evolutionary theory perhaps, but also for much that is going on these days in human biomedical (and also evolutionary) genetics, where causal connections between genomic genotypes and phenotypes are the interest.  In evolution, selection only works on what is inherited, mainly genotypes, but if causation is too complex, the individual genotype components have little net causal effect and as a result are hardly 'seen' by selection, and evolve largely by chance.  That's important because it's very different from Darwin's notions and the widespread idea that evolution is causally rather simple or even deterministic at the gene level.

Put another way, genomic causation evolved via the evolutionary process.  If natural selection didn't or couldn't refine causation to a few strong-effect genes, that is, to make it highly deterministic at the individual gene level, then biomedical prediction from genome sequences won't work very effectively.  This is especially true for traits, disease or otherwise, that are heavily affected by the environment (as most are) or for late-onset traits that were hardly present in the past or arose post-reproductively and hence didn't affect reproductive fitness and are not really 'specified' by genes.

There was considerable genomic variation between the authors' two replicate yeast experiments.  As one might say, meta-analysis would have some troubles here.  Likewise, from cell lineage to cell lineage, different sets of mutations were responsible for the fitness of the lineage in this controlled, fixed environment. This means that even in this very simplified set-up, genomic causation was very complex.  No 'precise' yeastomic prognostication!

In real biological history, even for yeast and much more so for sexually reproducing species in variable environments, selection has never been unitary or fixed, and genomes much more complex. Human populations have been until very recently very much smaller than 10^8 in the yeast experiments, and recent population expansion will make the number of low-frequency variants much greater, and with recombination, vastly more genomically unique.

The bottom line here is that our traits should be much less predictable from genotypes than traits in yeast. We have not reached, nor did our ancestors ever reach, the kind of fitness equilibrium reached in the yeast study under controlled selection, and fixed environments.

Somatic mutation
The authors also compare the large numbers of cells whose evolution they were able to follow with their barcode-tagging method, to the evolution of genetic variation in cancer and microbial infections, where there are even larger numbers of cells in an affected person and, importantly, clones expanding because of advantageous mutations. From the yeast results, these clonal advantages may not generally be due to one or two specific mutations (with perhaps, hopefully, exceptions when chemotherapy or antibiotics exert far stronger selection than was imposed in the yeast experiment). But the general complexity of such clonal expansions present major challenges, because they may end up with descendant branches distributed throughout the body where even in principle the responsible variation can't be directly assessed.

But the implications go far beyond cancer.  As we've recently posted, cancer is a clear but perhaps only a single manifestation of a more general phenotypic relevance of the accumulation of somatic mutations, that occur in body cells during life and can in aggregate have systemic or organismal-level implications.  The older we get the more likely we are to generate such clones, all over the body, and it seems likely that they can become manifest not just as individually ill-behaving cells, but as disease for the whole person.

But it's not just late onset implications that the yeast work may forebode.  There are already huge numbers of cells in the early embryo and fetus whose even huger descendant clades of cells during life grow many, many fold by adulthood.  There is no reason not to expect that each of us will carry clades that include differently-than-normal functioning cells in our tissues.  Let age, environmental exposure, and further mutations add to this and disease or age-related degeneration can result.  Yet none of this can be detected in the usual individual's 'genome' as currently viewed.  This is a potentially important fact that, for practical reasons or what one might call reasons of convenience, is ignored in the wealth of mega-sequencing projects being lobbied for based on genome sequencing (precision prediction being the most egregious claim).

So a bit of brewer's yeast may be telling us a lot--including a lot that we don't want to hear. Inconvenient facts can be dismissed.  Oh, well, that's just yeast!  They evolve differently!  That was just a lab experiment!  Brewers and bakers won't even care!

So let's just ignore it, as if it only applies to those rarefied yeast biologists.  Eat, drink, and be merry!

Wednesday, February 25, 2015

Survival of the safest: Darwinian conservatism, not derring-do

A mantra for many in life science is 'survival of the fittest'.  This phrase, one Darwin liked and used many times after he saw its use by Herbert Spencer, reflects Darwin's view of life as a relentlessly competitive phenomenon.  To Darwin, life was an unending struggle for survival (and reproduction) among individuals in every species all the time.  Natural selection, a relentless force like Newtonian gravity, always identified the 'fittest', weeding out the others.

Darwin's objective was to show how new characters could arise, that were suited--'fitted'--to their environment, without the intervention of God via special creation events.  Because organisms, all and always, were struggling against each other for limited resources, they 'tried' (via their inherited genomic drivers) to be better, different, more exploitive of environmental opportunities than their fellows.  Dare to be different!

But in perhaps fundamental ways, Darwin had it very wrong, perhaps inverted from what is really going on.  We know this from the analysis of genomes, the presumed source of all evolutionary evidence, since everything that's inherited goes back, at least indirectly and usually directly, to information carried in DNA.

When DNA sequences are compared within or between species, there are segments that are seen to have very little variation among the sequences, and segments with much more variation.  Now we have learned how to identify truly functional parts like coding exons, transcription start sites, introns, promoter and some regulatory regions, functional RNAs (like tRNA, rRNA and so on), telomeres, and so on.  And we have also identified many parts (the majority, actually) that has far less obvious or strong function, if indeed any function at all.  So what do we see?

The clear, consistent pattern is that the more strongly functional, the more highly conserved (there are a few exceptions, like sensory system genes in the olfactory and immune system, but even their variation proves the rule).  The less, or non-functional regions vary much more, both within and between species.
Herd of dairy goats, Polymeadows Farm; photo A Buchanan

This has been seen so consistently, that for many purposes (like the ENCODE project to characterize all DNA elements) sequence conservation is the very definition of biological function.  The reason is that evolution conserves function but doesn't care about bits that have no function.  One can quibble about the details, but the main gist of the message seems unequivocally correct.  But this now near-dogmatic principle has some little-digested implications.

Darwinian evolution
The problem Darwin wanted to solve was to explain the differences among species, and the way they were suited to their ways of life, in terms of historical processes rather than Divine creation.  He had a deep sense of geological change and biogeography from his trip on the Beagle, that showed the evidence of local relationships that suggested common ancestry.  And then he had an idea of a law of Nature, an ineluctable force-like process of adaptive change, the way gravity is a force, that would gradually form the kinds of differences that characterized species.

'Natural selection' was the name he gave to that force.  And because it was force, like the way gravity is a force, it could detect the tiniest differences among competing organisms and favor them to produce the next generation.

The idea is that species always over-reproduce relative to their resources (an idea that was already 'in the air' in Britain at the time), and struggle to obtain what become limited resources.  Because of inherited variation, the individuals with the best genotype (to use our term for it) reproduced, their poor lesser peers fell to what Tennyson would call 'Nature red in tooth and claw'.

Darwin's idea was that selection always favored the innovator.  Relentless striving to be different from the herd, to get the scarce food or mate supply.  As the late thinker Leigh Van Valen suggested, the Darwinian struggle was like the Red Queen in Alice and Wonderland--always running as fast as she could, but never getting ahead because the competition was always trying to out-do you with their own adaptations.

This is so entrenched in the biological and evolutionary literature, that it may be surprising to realize how different from what we see in the actual data--the genetic data, our most precise indicator--about how evolution works.

Or is this Darwinian?
What we actually see in the genetic data is not the kind of chaotic variation that an intense, force-like, relentless struggle to be better than your peers would lead us to expect.  Instead, what we see is what can only be called herd behavior at the genome level.  Our ancestors did, and our contemporaries do, their very best to stay with the herd.  The high conservation of functional DNA sequence suggests that mutational variation is mainly harmful to fitness, that what selection really favors is conformism. Don't be very different, or you'll get pruned away from your species' posterity!

Herd of flamingos, the Camargue, France; A Buchanan
Instead of 'survival of the fittest', what is by far mainly going on is 'survival of the safest'.  Stay with the mean.  Why is that?  A standard answer that is likely accurate, is that we today are the product of a long past in which the traits we bear were able to survive.  We're very complex organisms, so what evolution hath joined, let no one put asunder except at their peril!

Survival of the safest might suggest that there is no innovation, which is clearly not correct, since different species have different adaptations--fish swim, cats eat meat, bats fly, we write blog posts. So clearly differences do arise and have been favored regularly in the past.  However, that seems inconsistent with the high level of genomic conservatism that is so predictably identified.

One way, perhaps the major way, that these apparent contradictions are reconciled is this:  As Darwin stressed repeatedly, evolutionary change is very, creepingly slow.  That means that either there are occasional short bursts of rapid, major change, brought about by largely catastrophic changes in circumstances, or, only a very minor 'ooze' of the distribution of traits occurs in some favored direction from one generation to the next.  This is imperceptibly slow at any given time, because being near the mean is still the safest place to be.  Just be a tiny bit different.

Survival of the safest is Darwinian in that it is a form of natural selection.  Indeed, it is a lot more Darwinian than Darwin was himself.  Selection is more probabilistic and less force-like than he thought (he lived still in Newton's shadow), but it is always at work as he said it was.  It's just that it's mainly at work removing rather than favoring what is different.  Every geneticist knows this, but it is far from thoroughly integrated into the common view of evolution even by professionals.

If a trait were being strongly driven by selection in some new direction all the time, as in the more exclusive connotation of survival of the fittest, we might expect only a few variants with strong effect in the favored direction would be contributing to its newly adaptive instances.  Mapping the variation in the trait would perhaps yield a rather simple genetic causal picture as a result.

But if a trait is being roughly maintained, by survival of the safest, pruning away serious deviants, then any genotypes that are consistent with being somewhere near the average can stay around, with individual variants coming and going by chance (genetic drift).  Variants conferring trait values too far from the mean are pruned by selection.  But most variants can hang around.  Mapping would reveal very large numbers of contributing variants across the genome.  And there would not be precise predictability from genotype to phenotype.

This is what we see in biology.

Survival of the safest in daily life, too
'Survival of the safest' thus seems to be a better metaphor for adaptive biological evolution.  But if you think about it you'll see that we see much of the same regularly in most aspects of our society. We may say heady things like 'dare to be different!', but those who dare to be very different are quickly punished by being ignored or directly slapped down.  This is true through history. It's the general fact in religion, government, social behavior.  And it's true in science, too.  Business as usual is safe, real innovation is a threat to the established.  We see the press of society to claim to be different, but not really to be different.  Innovation mostly means incremental change trumpeted with exaggerated verbiage.  We may even think we want major, rapid change, but emotionally we shy from it, and feel too nervous about what it might mean for our own current state.

Organizations and sociocultural and political systems are very slow to change. or they change in herd-like fashion.  This is true even in realms, like the business world, where one often hears a rather self-satisfied pronouncement that allowing free-market Darwinian competition is the way to get innovation. Innovation is often claimed, but much less often really major.  It's true in science, too: everyone is playing grantsmanship, to seem different to draw attention or funds, in this case, but you dare to be very different at your peril, lest reviewers suspect, distrust, can't grasp, or are jealous of your idea.  Some rapid change may occur, but it's not so common relative to the inertia of survival of the safest.

That is what we see in society.

Tuesday, February 24, 2015

Causation revisited again

A paper* published recently in The Medical Journal of the Islamic Republic of Iran ("X-ray radiation and the risk of multiple sclerosis: Do the site and dose of exposure matter?" Motamed et al.) explores the possibility that X-rays are a risk factor for multiple sclerosis (MS).  (Do we routinely read this journal?  No.  Ken sent me a pdf of the paper, and when I asked him where he'd gotten it, he said he thought I'd sent it to him.  Which I had not.  On looking back at the email, he finds that it contained no actual message, just the pdf, and not even an identifiable sender.  Creepy spam? I guess we'll find out.  But until our computers are taken over by bots, despite its iffy provenience, the paper does bring up some interesting questions.)

From the paper abstract:
Methods: This case-control study was conducted on 150 individuals including 65 MS patients and 85 age- and sex-matched healthy controls enrolled using non-probability convenient sampling. Any history of previous Xray radiation consisted of job-related X-ray exposure, radiotherapy, radiographic evaluations including chest Xray, lumbosacral X-ray, skull X-ray, paranasal sinuses (PNS) X-ray, gastrointestinal (GI) series, foot X-ray and brain CT scanning were recorded and compared between two groups. Statistical analysis was performed using independent t test, Chi square and receiver operating characteristics (ROC) curve methods through SPSS software. 
Results: History of both diagnostic [OR=3.06 (95% CI: 1.32-7.06)] and therapeutic [OR=7.54 (95% CI: 1.59-35.76) X-ray radiations were significantly higher among MS group. Mean number of skull X-rays [0.4 (SD=0.6) vs. 0.1 (SD=0.3), p=0.004] and brain CT scanning [0.9 (SD=0.8) vs. 0.5 (SD=0.7), p=0.005] was higher in MS group as well as mean of the cumulative X-ray radiation dosage [1.84 (SD=1.70) mSv vs. 1.11 (SD=1.54) mSv; p=0.008].
So, it was a very small study, but the odds ratios were quite significant, particularly for therapeutic X-ray, for which dosage is likely to be higher than for diagnostic X-ray.


Chest x-ray; Wikipedia

And this isn't the only study, in fact, that has found an association between X-ray and MS.  Axelson et al. find a similar link in Sweden, described here, also in a very small study.  But, this about exhausts the reports of such a link.  The problem is that the risk of MS is small (the National Multiple Sclerosis Society estimates that there are 400,000 people in the US with MS, or about 1/1000) relative to the number of people getting X-ray, therapeutic or diagnostic.  This means that even if X-ray is causal, and the odds ratios (relative risk) in these small studies fairly large, the actual (absolute) risk is minuscule.

The cause of MS is unknown.  Many hypotheses have been considered -- it may be immune-related, or viral, or genetic, or perhaps environmental, and some believe lack of vitamin D is a prime candidate.  But as with other complex diseases, with the kind of varying and complex phenotype that is seen with MS, it's possible that there are numerous causes, and/or numerous triggers, rather than a single one.  So, if X-ray really is causal, perhaps it causes some tissue irritation that stimulates immune response, triggering some over-response that contributes to MS risk.  Thus, it's conceivably possible that X-ray is in fact contributory in some cases.  But one can imagine many such explanations.

But this again raises a larger question, one we've been blogging about off and on, well, forever, but recently, including last week, and again yesterday with respect to the new dietary recommendations, that no longer include cautions against eating foods high in cholesterol.  Why is it so hard to determine the cause of so many diseases?  Why don't we yet know the cause of MS, or heart disease, or obesity, or many other common diseases?  Essentially, it comes down to the fact that our methods for determining causation just aren't good enough when every case is different.

In the near future, we'll write about this issue in the context of how epidemiology is done these days.

--------------
*Here's the link, if you, too, want to chance it.  http://mjiri.iums.ac.ir/browse.php?a_code=A-10-1-758&slc_lang=en&sid=1&sw=sclerosis

Monday, February 23, 2015

When the methodology fails

Aop-ed piece by Nina Teicholz in Friday's NYTimes lays it on the line, chastising the government for its regular bulletins on dietary advice that, for 50 or so years have altered what we eat, what we fear to eat, and what the risks are.  Now, new studies tell us that what was bad is good and what was good is bad, and that the prior half-century of studies were wrong.  We've eliminated fats and cholesterol, and replaced them with carbohydrates, but, as Teicholz writes,
...recent science has increasingly shown that a high-carb diet rich in sugar and refined grains increases the risk of obesity, diabetes and heart disease — much more so than a diet high in fat and cholesterol.
But why should we believe these new studies?  Teicholz basically takes the underlying methodology to task, and yet she has written a book recommending that we eat more fats (“The Big Fat Surprise: Why Butter, Meat and Cheese Belong in a Healthy Diet"), but those recommendations are based on the very same faulty methodology as the recommendations with which she, and the current USDA advisory committee, find fault.

Embrace the fat! (Wikipedia)

The same, almost exactly the same, critiques are earned by many of the 'big data' genomics studies (and other long-term go-not-very-far megaprojects).  It is the statistical correlation methodology.  When many factors are studied at once (perhaps properly since many factors, genetic and environmental, are responsible for health or other traits), we can't expect simple answers.  We can't expect correlation to imply causation.  We can't expect replication.  We can't predict the risk factors that people, for whom risk advice is based on such studies, will face in the future.

The real conclusion is to shut down the nutrition megaprojects at Harvard (singled out by the op-ed) and the other genetics and public health departments that have been running them for decades, and do something different.  The megaprojects have become part of the entrenched System, with little or no real accountability.

Pulling the plug would be a major acknowledgment of failure, both by the feds for what they funded, the program officers for defending weak portfolios and their budgets, the universities defending their overhead and prestige projects and, of course, the investigators who are either simply unable to recognize what they're doing, or too dishonest and self-protecting to come clean about it.  And then they and their students could go on to do something actually productive.

Of course such a multi-million dollar threat will be resisted, and that's why the usual answer to the kinds of conflicting, confusing reports that so often come out of these megaprojects is to increase their size, length and, geez, what a surprise!, their cost.  To keep funding the same investigators and their proteg├ęs.  This is only to be expected, and many people's jobs are covered by the relevant grants, a genuine concern.  However, research projects are not supposed to be part of a welfare system, but to solve real problems.  And the same peoples' skills could be put to better use, addressing real problems in ways that might be more effective and accountable.

And we used to laugh at the Soviets' entrenched, never successful, Five Year Plans!

It is a public misappropriation that is taking place.  Yes, there are health problems we wish to avoid, and government and universities are set up to identify them and recommend changes.  But, for most of today's common chronic diseases, lifestyle changes would largely do the trick.

But then, that would just let people live longer to get diseases that might be worse, even if at older ages.  And meanwhile we aren't putting on a full court press for things that really are genetic, or really do have identifiable life-style causes.

Much of this research is being done at taxpayer expense.  We should let the people keep their money, or we should spend it more effectively.  We won't be able to do the latter until we admit, formally and fully, that we have a problem.  Given vested and entrenched interests, getting that to happen is a very hard trick to pull off.

Monday, February 16, 2015

Occasionality (vs Probability)?

In genetics and epidemiology, we want to explain the causes of diseases, and we want to be able to predict who will get them. We generally do this statistically, by working out the probability that we are right in our guesses about alleles or environmental causal factors.  But this is problematic for a number of reasons.

The concept of probability and the term itself have always been elusive or variable in meaning.  But they refer to collections of objects and the relative proportions of some specified component, or alternatively to the distribution of outcomes of repeated identical trials.  They work great for truly repeatable phenomena such as flipping coins or rolling dice or sampling colored balls from the proverbial urn (actually, there are problems even here, of both practical and theoretical sorts, but they are minor relative to the point we wish to make).

Dice; Wikipedia


In discussions of the phenomena pertaining to evolutionary genetics, 'probability' arises all the time.  We have genetic drift--the 'probability' of change in frequency of a given allele in a population of specified size; natural selection and evolutionary fitness--the relative probability of the bearer of a given genotype to reproduce relative to the probability of alternative genotypes (adaptive fitness); the probability of migration and hence gene flow, and much more.

The same sorts of things arise in biomedical genetics: Mendelian inheritance--the probability of a given allele being transmitted to a given offspring; the relative probability of a disease arising in a person of a given genotype; the probability of lung cancer in a smoker (per cigarette consumed, etc.); the probability of having a given stature or blood pressure in a person of a given genotype, and much more.

Such things are as routinely discussed as the daily news, or what some guest said on The Daily Show. But there is a problem.  The meaning of the term is unclear, often to the user as well as hearer. The term is deeply built into many areas of evolutionary and causative genetics, our means of drawing conclusions from both observational and experimental data.  But upon closer scrutiny, the term is  being used based on highly questionable or hopeful assumptions that at best only apply to a subset of situations.

Replicability and probability: p-eeing into the wind
The standard way to represent a probability is by a symbol, often the letter p.  Once you have identified this value, it is easy to write it into various equations, like, for example, the probability of observing 3 Heads in a row is p^3 (p cubed) where p is the probability of Heads on any given flip. Or, if you assume that your data involve sampling from the same urn filled with certain proportion of red and blue balls (or, say, of diabetics or tall people in a sample from a population of people with a given genotype) and the probability of getting your result just by chance (that is, no genetic causation involved),  p is less than 0.05, then you declare to the world that something unusual is going on--that is, your research has found something (such as that the genotype is a causal factor for diabetes or tall stature).  But if your actual set-up is not something simple like balls in an urn, or dice—that sort of thing, then what is it that is really happening?  How reliable or interpretable are these probabilities?

The assumed sort of regular repeatable condition, in the real evolutionary and biomedical world where we live, is usually a fiction!  That doesn't mean there is no truth involved.  It means that we think there is some truth in how we've organized our description of what's going on, and in a somewhat circular way we're using our data to check that.  Unfortunately, we usually don't know even what we think is going on in a rigorous enough way for the p's we employ to be accurate, or to put it more fairly, they are inaccurate to an unknown extent.  Is that extent even knowable in principle?

In the areas of biology and health discussed here, we are usually far from having such knowledge. We use 'p' values or their statistical equivalents all the time, both for the processes (e.g., risk of some disease or outcome) or unusualness in support of our causal assertion (e.g., statistical significance).

In most or perhaps every case, we are making strong underlying assumptions about the real world that is our objective to understand.  That assumption is one of replicability--that is, that we are dealing with repeatable events, often that there is some ‘random’ aspect of causation (itself usually a vague assumption not seriously explained).  That in turn implies that there truly exists some p-values or probability distributions underlying what we are studying.  Repeatability is a basic assumption, what one might even call a tenet about the nature of the organization of the world we are studying.  

In what sense do such distributions or probabilities actually exist?  Is the world of our interest organized, symmetric, and repeatable enough for the models being tested to be realistic? If the cosmos were purely deterministic (e.g., no truly probabilistic quantum phenomena), then we would always be sampling from worlds of proportions and our probability values would reflect repeated sampling events ('random' assuming we really knew what that meant in practice).

But suppose to the contrary that to an important extent each event is unique, even if perhaps wholly predictable if had we enough information, and that it is literally impossible for us to be both in the world and have enough information about the world in order to make such predictions.  Perhaps either truly, or pragmatically, the 'probabilities' on which so much of what we are about in modern genetics, and, of course, the promises being made, simply don't exist.  To the extent that's true, the edifice of probabilistic science and statistical inference are mistaken or misleading, to an unknown and probably unknowable extent.

Occasionality: a more appropriate alternative concept--where there's no oh!
When many factors contribute to some measured event, and these are either not all known or measured, or in non-repeatable combinations, or not all always present, so that each instance of the event is due to unique context-dependent combination, we can say that it is an ‘occasional’ result.  In the usual parlance, the event occasionally happens and each time the conditions may or may not be similar.  This is occasionality rather than probability, and there may not be any 'o-value' that we can assign to the event.

This is, in fact, what we see.  Of course, regular processes occur all around us, and our event will involve some regular processes, just not in a way to which probability values can be assigned.  That is, the occasionality of an event is not an invocation of mystic or immaterial causation.  The word merely indicates that instances of the event are individually unique to an extent not appropriately measured, or not measured with knowable accuracy or approximation, by probabilistic statistical (or tractable deterministic) approaches.  The assumption that the observations reflect an underlying repeated or repeatable process is inaccurate to an extent as to undermine the idea of estimation and prediction upon which statistical and probabilistic concepts are based.  The extent of that inaccuracy is itself unknown or even unknowable.

There are clearly genetic causal events that are identifiable and, while imperfect because of measurement noise and other unmeasured factors, sufficiently repeatable for practical understanding in the usual way and even treated with standard probability concepts.  Some variants in the CFTR gene and cystic fibrosis fall into that category.  Enough is known of the function of the gene and hence of the causal mechanism of the known allele that screening or interventions need not take into account other contextual factors that may contribute to pathogenesis but in dismissible minor ways. But this seems to be the exception rather than the rule.  Based on present knowledge, I would suggest that that rule is occasionality.

One problem clearly is definitional.  Where causation has traditionally been assigned to some specific outcome, we now realize that outcomes, like causes, constitute a spectrum or syndrome of effects, so that we have as much variation in 'the' outcome as in the input variables.  Clearly things are not working!

Occasionality can be vaguely viewed as something like a rubbery lattice of non-rigid interacting factors, with some connections and factors (nodes) missing or added from time to time without known (or knowable?) causal processes (e.g., what new mutations might lead to part of the genome to contribute to some spectrum of phenotype outcomes?), connecting rungs being broken or new ones added as the lattice jostles down a roiling stream in and around obstacles.  Particular vortices, or outcomes, a set of vertices (traits) that may or may not bob, occasionally, up to the surface individually or in different combinations to be measured as 'outcomes'.  This is at least an attempt at a heuristic way to envision the elusive fish the biological community are trying to land.

Metaphysics and physics envy
It has often been observed that biology's love affair with the kinds of mathematical/statistical approaches being taken is a form of physics envy.  It's an aspect of the belief system that pervades both evolutionary and genetic thinking.  It avoids the somewhat metaphysical reality, that we know life involves emergent properties that are not enumerable in ways ordinary physical and chemical systems seem (at least to outsiders like us, and at the macro-scale) to involve.  If what is wrong or missing (if anything) were clear, these areas of the life sciences would be doing things differently.

This may sound far too vague and airy, but the evidence is all around us.  Last week, as we wrote about here, even 40 or more years of nutritional studies, very sophisticated and extensive, are reported to have been 'wrong' in regard to the risk of dietary cholesterol, and salt consumption.  'Wrong' means not supported by recent research.  The nature of even of blood-levels of many factors like these also seems ephemeral, that is, findings  of one study are contradicted by another study.  Which (if any) are to be believed?  Why do we think our new studies or methods obsolesce the older ones, when in essence there aren't very substantial differences?  Why haven't we got solid, replicable, clear-cut causal understanding?

The difference between occasionality and probability is somewhat like the difference between continuous and discontinuous (quantitative vs qualitative) variation.  The challenge is to understand and make inferences about the latter when we don’t have underlying analytic (mathematical function based) theory to apply.  We don’t want to say an effect has no cause (unless we are involved in quantum mechanics, and even that's debated!), but we do need to understand unique combinatorial causation in ways we currently don’t.  This will have to be mixed with the statistical (probabilistic) aspects we can't avoid, such as sampling and measurement effects.

There are many reasons to dismiss such a term or concept as occasionality.  First, because it is not operationalized: we don’t really know what to do about it.  Second, there is a tendency to reject some new term that is invoked in a critical vein.  Thirdly, the practical reason is that even if the idea is correct, if it were acknowledged too fully it would undermine the business that most in this area of science are about, and certainly would be resisted.  But if we were to force ourselves to ask what we would do if we knew the phenomena we want to understand were manifestations of occasionality rather than probability, it might force us to think and act differently.

Thursday, February 12, 2015

What's a 'healthy diet' anyway?

A nutrition advisory panel is convened by the US Department of Agriculture every five years to review recent research in the field, make sense of it, and offer recommendations about what Americans should be eating for optimal health.  Those food pyramids, MyPlate, decades of advice to limit our cholesterol, saturated fat and salt intake?  All the work of these panels of experts who scoured the data and told us what it meant.  And as a result, from the 1960's onward, good, conscientious people reduced their cholesterol intake, took the salt shaker off the table and the whole milk and butter out of the refrigerator, cattle were bred to be leaner, eggs were banned from breakfast, low-salt and low-fat processed foods appeared on the shelves, and heart disease death rates ... continued to fall.


Deaths due to diseases of the heart (United States: 1900–2006). Circulation.2010; 121: e46-e215

Now, according to numerous news accounts, including here at the Washington Post, after more than 50 years of anti-cholesterol, anti-fat expert advice, the current advisory panel is reportedly poised to recommend that we need no longer need to limit the amount of cholesterol we eat, nor worry about our salt intake.  Bring on the shrimp, the lobster, the eggs, the meat, banish the guilt!

Oh, wait.  Hold on a sec.  Let's go back to those falling heart disease rates.  I remember one of the first lectures I heard as a new grad student at the University of Texas School of Public Health in the late 1970s, given by heart disease epidemiologist Reuel Stallones, Dean of the school.  His point was that rates had been falling since the 1960's and epidemiologists had no idea why.  He systematically destroyed every argument then (and now) current that might explain the rise and fall of heart disease death rates -- changing diet, decreased smoking, de/increased exercise.

In fact, he made the same case in 1980 in Scientific American in an article called "The Rise and Fall of Ischemic Heart Disease".  (The terms ischemic heart disease, coronary heart disease, and arteriosclerotic heart disease are more or less interchangeable, according to Stallones.)
In the U.S. the death rates attributed to heart attack and other results of the obstruction of the arteries that nourish the heart have fallen since the 1960's. Why they have is not understood.
Here's Stallones' graph of heart disease death rates, from 1900 to 1980.  I can't explain why deaths are still rising in the above graph in the 1950's, but falling in this graph; presumably this has to do with differing classifications of deaths due to heart disease.  Anyway, here rates rose rapidly starting in 1920 and then began to fall in the 1950's, steeply in the 1960's.
Stallones, Scientific American 354(5); 53-59


Here's another graph showing the decline more or less in line with Stallones' data, from a 2000 paper in Circulation.

"Death rates for major cardiovascular diseases in the United States from 1900 to 1997. *Rates are age-adjusted to 2000 standard." Source: Circulation, Cooper et al., 2000

As Stallones wrote, "Plainly a sustained decline in the death rate for ischemic heart disease commands attention and calls for explanation." And, he backs up to ask not only why the fall, but why it was that heart disease mortality began to rise so quickly in 1920, particularly among men.  Whatever explains the decrease must also explain the increase.

Smoking began to rise after World War 1, which fits the rise in heart disease mortality, and began to decrease from the mid-1960's or so, which fits the decrease.  But such an explanation would assume no latency period between beginning to smoke and its effects on heart disease.  And, middle-aged men quit smoking at higher rates than women, and this is not, as Stallones said, in concordance with the pattern of decline in heart disease mortality.  Treatment isn't an effective explanation either, because incidence rates -- new cases -- followed the same pattern as mortality.

He concluded,
In summary, four major variables are known to be associated with the risk of ischemic heart disease in individuals. Among the four, hypertension does not fit the trend of the mortality from ische­mic heart disease at all; physical activity fits only the rising curve, serum choles­terol fits only the falling curve and only cigarette smoking fits both. In no case is the fit as precise as one would like. This raises doubt that any of the factors is a fully satisfactory explanation for the variation in mortality.
So, in sum, as of 1980 epidemiology had no explanation for the rise and fall of heart disease mortality rates.

And epidemiology still can't tell us what causes heart disease, or predict who'll get it.  So, apparently we'll soon be told that we no longer have to monitor our cholesterol intake, and there's a lot of talk about fat consumption not being linked with heart disease anymore, and it's not clear whether obesity or hypertension are actually causal.  At least smoking is probably still a problem.

It's clear from the rise and fall of death rates through the 20th century that genes aren't going to be the major explanation because genes can't explain the spike in the 1920's and the fall 40 years later.  That experience also makes it clear that we can't predict environmental changes (or, often, even figure out what they were in hindsight) that might be associated with risk, and thus we aren't going to be able to predict the future, despite the claims of precision medicine advocates.

Stallones suggested that heart disease mortality data might indicate a single environmental cause to explain the rise and fall of death rates, but found reasons to argue against each of the most obvious ones.  Could the cause have been inflammatory?  If so, that would reinforce the idea that predicting future environments and causes is not going to be possible.

And, if there was a single cause, it's curious that our reductionist approaches, with large carefully designed samples and sophisticated statistical analysis, were unable to identify it, because that's what they are widely thought to be best at.  This makes it more likely that heart disease in populations has multiple causes.  And in fact every heart attack is unique, because no two people eat the same things, do the same amount of exercise, suffer the same infectious diseases, and so on. So maybe the very word 'cause', and the very approach (statistical), both of which assume some regular, replicability properties, are not being appropriately conceived.  This is a subject we'll discuss next time.....

"Experts" responding to the coming cholesterol recommendations have said that we still need to eat a healthy diet.  But when we still have no idea what's unhealthy, it's hard to know what is.

Monday, February 9, 2015

Simulation: surprisingly common!

The other day we did a post suggesting that given the commitment to a mega-genomic endeavor based on the belief that genomes are predictive entities about our individual lives generally, we should explore the the likely causal landscape that underlies the assumption, with tools including research-based computer simulation.  Simulation can show how genomic causation can most clearly be assessed from sequence data.  By comparison to the empirical project itself, simulation is very inexpensive and trivially fast.  Research simulation has a long history of various sorts, and is a tool of choice in many serious sciences.  Research simulations are not just 'toy model' curiosities, but are real science, because the conclusions and predictions can be tested empirically.

Still, many feel free to dismiss simulations, as we discussed. It seems in a sense too easy--too easy to cook up a simulation to prove what you want to prove, too easy to dismiss the results as unrelated to reality.  But what alternatives do we have?

First, roughly speaking, in the physical sciences there exists very powerful theory based on well-established principles, or 'laws'.  Laws of gravity or motion are classic examples.  A key component of this theory, is replicability.  In biology, our theory of evolution and hence of genomic causation is far less developed in terms of, yes, its 'precision'.  Or perhaps it's more accurate to say that as currently understood, genetic theory is fine but we are inaptly applying statistical methods designed for replicable phenomena for things that are not replicable as the methods assume (e.g., every organism is different).  In the physical sciences to a great extent, the theory comes externally to a new situation under study, that is, is already established and being applied from the outside to the new situation.  Newton's laws, the same laws, very specifically help us predict rates of fall here on earth, and also how to land a robot on a comet far away in the solar system, or even study the relative motion of remote galaxies.

By contrast, evolution is a process of differentiation, not similarity. It has its laws that may be universal on earth, and in their own way comparably powerful, but that are very vaguely general: evolution is a process of ad hoc change based on local genomic variation and local circumstances--and luck.  They do not give us specific, much less precise 'equations' for describing what is happening.

We usually must approach evolution and genomic causation from an internal comparison point of view.  We evaluate data by comparing our empirical samples to what we'd get if the experiment were repeatable and nothing systematic is going on ('null' hypothesis) or if it's more likely that something we specify is (alternative hypothesis).  That is, we use things like significance tests, and make a subjective judgment about the results.  Or, we construct some formulation of what we think is going on, like a regression equation Y = a + bX, where Y is our outcome and X is some measured variable, a is an assumed constant and b is an assumed relative-effect constant per unit of X.  This just illustrates what we have a huge array of versions of.

But these 'models' are almost always very generic (in part because we have good ready-made algorithms or even programs to estimate a and b). We make some subjective judgment about our estimate of the parameters of this 'model' such as that our a and b estimates are correct to a known extent and are in fact causal factors. One might estimate the value of a or b to be zero, and thus to declare on the basis of some such subjective judgment, that the factor is not involved in the phenomenon.  Such applied statistical approaches are routine in our areas of science, and there are many versions of the basic idea.  They are these days often, if informally and especially in terms of policy decisions, taken to be true parameters of nature.  For example, as in this simplistic equation, the assumption is made that the effects of a and b on Y are linear.  There is usually no substantial justification for that assumption.

In a fundamental sense, statistical analysis itself is computer simulation!  It is not the testing of an a priori theory against data, either to estimate the values of things involved or to test the reality of the theory. The models tested rarely are intended to specify the actual process going on, or are purposely ignoring other factors that may be involved (or, of course, are ignored because they aren't known of).

Proper research simulation goes through the same sort of epistemological process.  It make some very specific assumptions about how things are, generates some consequent results, which are then compared to real data.  If the fit is good then, as with regular statistical analysis, the simulated model is taken to be informative.  If the fit is not good then, as with regular statistical analysis, the simulation is adjusted (analogous to adding or removing terms, or differently estimate parameters, in a statistical 'model').  Simulation models are not constrained in advance--you can simulate whatever you specify.  But neither are statistical models--you can data-fit whatever 'model' you specify.

There are lots of simulation efforts going on all the time.  Many are using the approach to try to understand the complexity of gene-interaction systems in experimental settings.  There is much of this also in trying to understand details such as DNA biochemistry, protein folding, and the like. There is much simulation in population ecology, and in microbiology (including infectious disease dynamics).  I personally think (because it's what I'm particularly interested in) that not nearly enough is being done to try to understand genetic epidemiology and the evolution of genomically complex traits.  To a great extent, the reason in my oft-expressed opinion is the drive to keep the support for very large-scale and/or long-term studies, given the promises that have been being made of medical miracles from genetic screening.  Those reasons are political, not scientific.

Analytic vs Descriptive theory
I think that the best way to get closer to the truth in any science is to have a more specific analytic theory, that describes the actual causal factors and processes relative to the nature of what you're trying to understand.  Such explanations begin with first principles and make specific predictions, as contrasted to the kind of generic descriptive theory of statistical models.  Both approaches are speculative until tested, but generic statistical descriptions generally do not explain the data, and hence have to make major assumptions when extrapolating retro-fitted statistical estimates to prospective prediction.  The latter is the goal of analytic theory, and certainly what is being grandly promised.  Here, 'analytic' can include probabilistic factors including, of course, the pragmatic ones of measurement and sampling errors, etc.  It is an open question whether the nature of life inherently will thwart both of these, leaving us with the challenge to at least optimize the way we use genomic data.

It is not clear if we have current theory that is roughly as good as it gets in regard to explaining evolutionary histories or genomic predictive power.  Life certainly did evolve, and genomes do affect traits in individuals, but how well those phenomena can be predicted by enumerative approaches to variation in DNA sequences (not to mention other factors we currently lump as 'environment') based on classic assumptions of replicability, is far from clear.  At least, at present it seems that as a general rule, with exceptions that can't easily be predicted but can be understood (such as major single-gene diseases or traits with strong effects of the sort Mendel studied in peas), we do not in fact have precise predictive power.  Whether or when we can achieve that remains to be seen.

There is one difference, however:  statistical models cannot be applied until you have gone to the extent and expense of collecting adequate amounts of data.  In truth, you cannot know in advance how much, or often even what type of, data you would need.  That is why, in essence, our collect-everything mentality is so prevalent.  By contrast, prospective research computer simulation can do its work fast, flexibly: it only has to sample electrons!

Denigrating computer simulation should not be done by anyone who is actually doing computer simulation, by calling their work by another name such as statistical modeling.  There is plenty of room for serious epistemological discussion about whether or how we are or aren't yet able to understand evolutionary and genomic process with sufficient accuracy.