Monday, 20 June 2016

Brexiteers

I have just received the EU Myth Buster from the Brexiteers. Firstly it still has the already proven wrong 350 million a week claim, as well as the nonsense about expansion to new members including Turkey. This is all scare mongering and the gall of the Brexiteers to claim it is Remain that are using economic fear to win the argument is just nonsense.

On the back the great and the good making their arguments. I wanted to go through them one at a time.

1) Sir Richard Dearlove - former chief of MI6.

"Brexit would bring about two potentially important security gains: the ability to dump the European Convention on Human Rights ... and more importantly, greater control over immigration from the European Union."

Sorry but this makes me very worried when a former head of the security service thinks that not having human rights is a good thing. This is the person who was in charge during 9/11 and then the dodgy dossier fiasco. Regarding his second statement, is the EU a source of terror threats to the UK? The 7/7 attacks were home grown and we do control immigration from outside the EU. He is also on record as stating that the media have exaggerated the threat from Islamic terrorists. So maybe he thinks the French farmers have been radicalised and are a threat?

2) Tim Martin - chairman of Weatherspoons

"The EU places tariffs on goods from outside the EU, which is bad for British shoppers and the developing world. And the EU forces us to charge VAT on goods, pushing up bills for working families."

By his argument leaving the EU and becoming one of those countries outside this is going to be good because now we are subject to those tariffs he is talking about. The levels of taxation are set by governments and they are to make sure that the finances of the country are in good order. The UK sets its own levels of VAT. These had to be increased because of the banking crisis not the EU. When we face tariffs because we are outside of the EU tax levels are likely to rise even more.

3) Nigel Lawson - former Chancellor of the Exchequer.

"As Chancellor, I became increasingly aware that, in economic terms, membership of the EU did us more harm than good. Outside the EU, we would prosper, we would be free and we would stand tall."

Sorry Nigel but this is pure rhetoric without a single shred of evidence or valid argument. He was last Chancellor in 1989 before the Berlin wall had fallen. The world has changed a bit since then and certainly the economy has changed beyond all recognition. Free and tall are not economic arguments, they are useful for boxers and cage fighters but they do not secure wealth for nations.

4) John Longworth Director of the British Chambers of Commerce.

"The EU interferes with UK firms and stacks the rules in favour of a select number of big businesses. It we Vote Leave jobs will be safer. We can have faster growth and greater prosperity in the future."

The EU spends a very large amount of its business spending on what are called SMEs and not the very large firms. Big business and big interests are very effective at lobbying the EU. My dad was a lobbyist for the Cattle Breeders Association and the UK farming industry which is hardly a big business. He would have never dreamed of voting out. The EU actually is more effective than the UK in dealing with the excesses of global business such as Microsoft and Google. These are companies that individual nations do not take on. In fact the UK government is noticeable in its reticence to get involved in taxing many of these businesses (Vodaphone is a high profile example) which get special tax deals that are certainly not given to small UK firms. Exactly the opposite to his claims is closer to reality.

5) Gisela Stuart Labour MP

"The rights we have won for British workers came from our Parliament, not the EU. The EU is run in the interests of the big corporations who spend billions lobbying to make it work for them."

I think that billions is an exaggeration. Even big business cannot spend that much on a Parliament that has so little legislative authority. I seem to remember the Conservatives opposing EU legislation precisely because it did protect worker's rights and asking for opt-out clauses. I seem to remember the social charter causing uproar from Mrs Thatcher, which does not agree with the MP's claims. It was only finally accepted in 1998 by a Labour government. I was also discussing this issue and where Deutsche Bank would move to if there was Brexit and the wife of one of their employees said not back to Frankfurt because they are only allowed to work a 37 hour week in Germany because of the EU legislation (that has not bee adopted by the UK) whereas her husband has a 67 hour week here.

It is amazing that in 5 quotes there is not a single good reason to Brexit. There is nothing there that is not fabricated, untrue or not pure fallacy. If that is the best they have got in their arguments then the Remain camp have very little to worry about.

Wednesday, 13 April 2016

Thoughts on Heritability

Using Waddington's idea of canalisation. This is the related to the width of the canalisation - this is the maximum variation between a pure in bred low phenotype and a pure in bred high phenotype. This is the variance in that characteristic which is attributable to the genes.

This is going to be very hard to measure unless you do it for a genuine population. If you have a non representative sample then that will have its own variability and the variation in that characteristic will depend on the state of all the different end points you are starting from in a normal mating population carrying all of its history and variation. Think of dogs the dog population includes all the breeds but the variation within breeds is very low, the same with horses but with humans you have Usain Bolt and me to compare for our 100 metres performance.

The width of the canals is important as it shows the plasticity in response to environment. How far does it need to be pushed to get someone outside the normal bounds. The genes make the landscape and they do not determine the outcome except in pure bred lines where the canals are very narrow. 

High heritability could actually signify large canals and low determinism because we are very far from a pure bred line and everyone is close to the middle of a large canal. But it could also indicate low variation and a multi-modal population - mixing of species, comparing apples and oranges. Conversely low heritability measures would mean that there is low variation with very narrow canals and it is very strongly inherited.

The concept of heritability implies that you can have a pure bred line for a specific phenotypic characteristic and so that characteristic must be inheritable. So my reason for questioning the paper rejecting Haidt and heritability is that why would his classifications be capable of being bred for? Can I breed someone who believes in justice over everything? Can I breed someone who finds disgust in the unclean their biggest driving force? Can any of these higher human constructs be embedded into genes or are they more accurately modelled at the meme/cultural evolutionary level? This is what I meant by there being strong determinism - that you could breed for it and that the genes would determine the effect. 

My opinion is that they are at the cultural evolutionary level and only the very simplest of behaviours is canalised at the genetic level. This would be probably things like higher level reasoning, personal identity, desire to reproduce, desire for satisfaction etc. The advantage of the cultural/meme level is that it is NOT Darwinian. It develops during lifetimes and is directly passed on to the next generation. It is Lamarckian and not wasteful random exploration. It builds on what we already have and is rapidly modified. It is much faster than genetic evolution. 

Sunday, 20 March 2016

Spectator Review of Not in Your Genes

This is the spectator review by a post-doc in psychology.


Now there are some good references in the beginning but sadly the complaints and understanding also decline as it goes on. In particular it descends into an ad hominem attack against the author and not his work. If you are convinced of your argument why go after James and his media appearances?

The first few references seem fine about birth order and the 10,000 hours effects. Then is goes down hill. So what if 80% of genes are expressed in the brain - this tells you nothing more than the 20% that are not cannot have any effect on the brain. Why does this mean anything? Lots are house keeping genes that just keep cells alive.

The ample evidence is a meta analysis of twin studies. The problem with twin studies if the author had bothered to check is that epidemiologists now suspect that they might be flawed as they have the same exposures and circumstances. I have two children who are not twins but they share lots of views and personality traits because of their upbringing not because of their genes. It is very hard unless you separate twins to actually get good evidence for genetic as opposed to environmental effects and even then epigenetic effects might be larger than genetic effects (these follow Lamarck and not Darwin - they are "Just So Story" modifications that are directly heritable)

I will need to look at GCTA and think about how to weight that evidence.

Then there is a look at GWAS which again the review author just talks about a review chapter. I have seen the people who lead some of the major GWAS studies of disease speak and there conclusions are that the effects of genes are SO SMALL THAT THEY CANNOT BE STATISTICALLY DETECTED. The effect sizes are tiny in diabetes, heart disease and lots of less complex phenomena than learning and personality. This is because it is gene interactions that produce the effects and not single genes. This talk was from the same lecturer who cast doubt on the twin studies. The review author might like to read Prof. David Clayton's work.

Then for basic statistics he cites wikipedia. While I like wikipedia this is all credibility lost. Power calculations are a tautology. You need to know effect size and population standard deviation in order to calculate the population size that you need to use to detect the effect. This is knowing the answer before you ask the question. The last references from the Journal of Irreproducible Results (Nature) is from the News and View section, an unrefereed section where the great and the good get to spout garbage on a weekly basis.

So apart from the possible evidence from GCTA there is actually as little to support the reviewers assertions of genetic links as there is for James support that there are no genetic links. The final result is a 0-0 draw.

Life imitates art: Captain America - Civil War

I have been reading twitter and the battle lines drawn between the genetics determine personality, ability and psychology and the opposing genetics tell you nothing camps. This is not the first time this battle has been fought. Last time we didn't even know very much about genes and we called it Eugenics. That was the Middle Class Victorian English man's attempt to protect his position in the world from the reality that he was nothing special. The second world war and the holocaust finished Eugenics as an acceptable position.

This new battle is much more dangerous especially in the current political climate and with current technology. There is every chance that we are going to up the scales to an existential level of threat and that will prove a Pyrrhic victory for whoever is left.

I am not a fan of Oliver James. I find his popular psychology grating and annoying and many of his arguments facile and unsupported by evidence. His most recent book has been attacked by almost all reviewers and a lot of people I follow on social media. It is titled "Not in Your Genes". The basic point is that nurture and not nature is responsible for all of your personality traits and abilities. You are not where you are because of genetics it is just because of up-bringing.

I have on my shelf a book written by Richard Lewontin, Steven Rose and Leon Kamin with a very similar title "Not in Our Genes" that makes very similar arguments. One of the criticisms of Oliver James is his lack of knowledge of genetics. It is impossible to doubt the genetic credentials of Lewontin. Lewontin et al's book was an attack on the genetic determinism and sociobiology of Dawkins and E.O. Wilson. Now I can seriously dispute the genetics ability of both Dawkins who has no idea about population genetics and molecular biology. As for Wilson his problem is that he studied the social insects and derives his rules for human social interaction from insect social interaction. While society is therefore clearly something that evolution has to discover (it is a universal in the language of Stewart and Cohen), he is mistaking analogy for homology. We do not have a common social history with the insects. These are two distinct evolutionary solutions to the same problem, and because of this knowledge of one CANNOT be directly extrapolated to the other.

James' view that genetic determinism is not a powerful driver is therefore not unsupported by people who do know their genetics as the critics might claim. Steven J Gould was another campaigner against the last vestiges of eugenics. Dan Graur is also particularly hostile in social media to psychologists claiming genetic effects or molecular biologists claiming function for all of the genome.

To understand it better we need to go back to the Victorian man who started this all. We need to think of Galton. As well as his work on genetics Galton is famous in statistics for discovering regression to the mean. The problem is that he and many other people failed to understand what it really means. It means that traits are not pure. It means that if someone is exceptional in something, then their progeny and likely to be closer to the mean. This is why so many properties in biology are normally distributed. The normal distribution is the error distribution for a complex system with many confounding variables. That sums up most of the properties of life, intelligence, success, height etc. If our genetic determinants were pure and with strong selection we would see divergence between members of the population until we would have speciation. We might have evolved the Morlocks and Eloi of H.G. Wells imagination. But we didn't. The strongest evidence that the genetic determinists are wrong is the diversity of humanity that still remains a single cohesive species.

If the determinists were right that genetics plays a major determining role then we could look at large genetic differences in humans and find differences. The largest difference is that between the genders where there is a Chromosome of differences and we could say that women and men (which the media seem to treat as distinct species) would have different personalities and levels of intelligence. This was an argument for a long period of time as Middle Class Victorian man tried to maintain himself at the top of the social tree. Now we know that if there are differences it is that women are more intelligent and have a better balanced personality than men but how much of this is social and experience and how much is genetics?

Next we could argue that populations that have been separated for long periods of time will have evolved different psychological and intellectual properties under genetic determinism. This leads us to the question of race as these are distinct genetic populations. Do we really want to go there and start asking those questions? Have we learnt nothing from two World Wars, endless genocides and the examples of peaceful migrations and settlement? There are no fundamental differences. Around the edges there are some but these are accidents of history and not universals of evolution.

This brings me back to the title. Science faces this civil war between the determinists and those who believe that nothing is set in stone. That is between the frightened Technologists like Tony Stark and those who have lived in a world where fascism was in the open and not cloaked in making X or Y nation great. So I am definitely on the side of the Captain and against rigid solutionism. This is why it feels to me like the Civil War. I am going to be on the opposite side of the argument to a lot of scientists who I respect and who I would have been allied to in the past.

Humanity works because of diversity and inclusiveness, for all of the bad that happens, what we do well out-weighs this and we must not let the cynics and the spreaders of fear win. We do not want to live under Ultron or Skynet.

Nature AND nurture play a part and the interaction between nature and nurture is the most difficult part to unravel. My intuition tells me that nurture is still the more significant of the two and that genetics at the level of personality and psychology plays a fairly small part. Nature and genetic evolution are very slow movers and slow to change but humanity has changed to be unrecognisable just in my life-time. This difference in time frames for me is the ultimate evidence that most of the arguments for genetic determinism will turn out to be wrong.

Saturday, 6 February 2016

Secondary Structure Prediction - Do neural networks add anything.

I began in bioinformatics when secondary structure prediction was considered an interesting problem. Now it is considered as "solved".

All of the early methods used sliding windows (from Scheraga) and statistical propensities of the amino acids (Chou and Fasman, GOR etc.). The big step forward was realising that predictions for alignments which accounts for positional variation was better than trying on only a single sequence. With the massive growth of the databases this has only got better.

Supposedly Neural Networks also helped improve predictions until we could do no better. The problem is are the NNs actually detecting any patterns that the straight linear statistics were not? They will probably help to find the amphipathic helices that the window methods will struggle with as this is a periodic pattern but they will not be able to deal with beta sheets as these are non local and exist outside of the windows. I had a student who constructed a neural network for prediction without any hidden layers and got the same prediction accuracy. This suggests that the neural networks are contributing nothing to the predictions.

We know that the codon distribution is optimised to make sure that mutations have a minimum effect of the resulting proteins (Baldi and  Brunak and Andreas Wagner).

We need to repeat the student's experiment and to check in the NNs actually make any difference. We also need to change the amino acid coding scheme to give realistic distances between amino acids and not Hamming distances.

Networks in proteins

A long time ago there was some interest in networks in proteins and a few articles were published about the non-bonded networks. These were interesting but they just looked at spheres of contacts around atoms to find a power law. The problem is that this power law arises from the cube relationship between number of atoms/residues and size - a protein approximates to a sphere. What would have been more interesting is to analyse the hydrogen bonding networks.


Friday, 5 February 2016

The open data bun fight, why experimentalists and analysts need to collaborate more.

I can understand why the people who collect the data want to protect their hard work and their grants against research parasites. The problem is that people like me who have spent their careers working with methods and thinking about the statistics are probably in a better position to carry out the analysis.

I have seen too many badly designed experiments that have wasted research funds because the experimentalists did not talk to the statisticians or the analysts. They end up with a badly designed experiment with so many flaws that it is practically useless.

Collaboration would be good for everyone. Yes it does mean one or two more names on the paper but the analysts don't want to be the submitting author so long as their contribution is recognised. Where being an analyst/bioinformatician/statistician will get annoyed is if someone treats them like a technician and not an equal party and collaborator. Analysts contribute equal skills to experimentalists, they are just different. Neither are more or less important and I have done both (I was an x-ray crystallographer solving protein structures as well as a molecular modeller using protein structures).