1 Comment

This is a brief review of the online course "Machine Learning Foundations: A Case Study Approach" from the University of Washington. I must disclose that although I worked through the examples in the course, I could not complete the assessments since I was not taking the course for credit.

The course was run by Carlos Guestrin, Amazon Professor of Machine Learning Computer Science and Engineering and Emily Fox, Amazon Professor of Machine Learning Statistics. Carlos is the co-founder and CEO of Dato, Inc (formerly GraphLab, Inc).

The ML (Machine Learning) blackbox used in the course was from GraphLab as well as the iPython Notebook programing language. These modules were downloaded to one's own computer which made it possible to play along and make coding mistakes. The fundamental modelling  design is illustrated in the diagram below -- using predicting house prices.


The examples used were:

  • Regression:  Predicting house prices
  • Classification:  Analyzing consumer sentiment
  • Clustering and Similarity:  Retrieving documents
  • Recommending Products
  • Deep Learning:  Searching for images

This course provided an excellent introduction into the world of machine learning. As the open source tools of ML become more sophisticated and easier to use, it opens the doors for anyone with an interest in data analysis or modelling for mind-blogging applications. Below is a video mashup of Carlos and Emily discussing the future of ML.

What is the bottom line about machine learning? It doesn't require any coding, it uses a tool-kit of statistical methods and the data is split into 'training data' and 'test data' where the training data is used to fit into a regression or a nearest-neighbor (or any other) statistic by iteration until there is convergence in the error and then this is evaluated by using the test data. The modelling is theory-free and only requires good data.

1 Comment

The DNA genome and machine learning would seem like an unlikely partnership since one is embedded in the biological world and the other in artificial intelligence. However, in recent times the human genome has been used as ‘training data’ for machine learning and has been able to predict the phenotype (in this case facial appearance) to a remarkable degree. (Photo from Riccardo Sabatini’s TED talk)


How was this done and how was machine learning used? Machine Learning (ML) is a "Field of study that gives computers the ability to learn without being explicitly programmed". My expertise in machine learning is somewhat limited and comes from a Coursera online course Machine Learning Foundations: A Case Study Approach from the University of Washington. My understanding of the process is illustrated in the diagram below (which may be only conceptually correct)

Machine Learning

The phenotypic characteristics of the subjects were codified and a random set was used along with their DNA sequences for ‘training’ the machine learning (ML) model.  A predictive model was made.  A set of ‘testing’ subjects was put through the model to evaluate the predictions. Further tweaking’ of the model was made through more iterations until some prediction endpoint was reached. This project involved the coordination of a large number of people working on different machine learning modules.

Watch the interesting TED talk by Riccardo Sabatini

At the end of Riccardo’s talk, he raised the issue that the human genome should be everyone's concern — philosophers, politicians, artists, scientists, business people and ordinary citizens. In the closing remarks of  my previous blog — I raised the concern: that since we have by and large eliminated ‘selection’ from the process of biological evolution,  we as a species shall continue to accumulate mutations in our genome — something that occurs on a daily basis. We potentially face an evolutionary dead end unless we are willing to intervene and correct these genetic mistakes. Also are we ready to grapple with the thorny problem of improving our genetic makeup?

As a species do we face an evolutionary dead end?.

This is the last blog posting about the Coursera sessions:  Genes and the Human Condition - (University of Maryland) . The first blog post dealt with some of the fundamental concepts and progressed through to the state-of-the-art technologies. This blog  highlights some of the genetic advances already made and their implications for society. Below is a  laundry list of the topics covered:

  • synthetic biology —> the concept
  • 1st transgenic example — human insulin from E. coli
  • ‘Pharming’ - using crops or animals for producing vaccines or drugs
  • aquaculture - AquaAdvantage salmon
  • golden rice - vitamin A and iron
  • BioBricks - standardised genetic components that can be linked together into new combinations
  • Craig Venter and Synthia — synthetic life
  • BioHackers - weekend workshops
  • gene therapy - replacing faulty genes with functional genes
  • CRISPR technology
  • germline therapy — benefits and the risks

A mash up of the lectures for this session --  Professor St. Leger

In the closing remarks of this course, there was a plea for the public to become more informed about biotechnology and genetics. Otherwise the fears of the new developments such as synthetic biology will block the benefits of such research for our future. We have eliminated the selection pressure on the many mutations that have accumulated in our genomes and will continue to do so. We face an evolutionary dead end if we fail to address the genetic consequences of no selection.

If you think you are being surrounded by idiots, you are probably right. A leading Stanford University scientist, Gerald Crabtree, would confirm your view. His idea is that since individuals are no longer exposed to nature’s raw selection mechanism on a daily basis that nearly all of us are genetically compromised compared to our ancestors of 3,000 years ago.

This and the previous blog are notes from the Coursera: ’Genes and the Human Condition’, University of Maryland, lectures on “My Genes Made Me Do It”. What follows is from lecture 5.

Crabtree gives the following example: "If a hunter-gatherer did not correctly conceive a solution to providing food and shelter probably died along with his or her progeny. Whereas a modern Wall Street executive that made a similar conceptual mistake would probably receive a substantial bonus and a be a more attractive mate". In other words, the consequences of being stupid were much worse in the Stone Age.

What is Self-domestication? - it is a process of transforming ourselves through self selection. What are some of the traits? We are selecting against aggression and favoring the juvenile behaviors  of trust, playfulness, and creativity. Both in brain size and physiology, we can be considered to be sexually mature baby chimps. This process is called paedomorphism, where the adult retains their infantile or juvenile features. Like most domesticated animals we have a smaller brain than our progenitors. Our brains have shrunk by about 20% in the last 10,000 years. The area that seems to have decreased in size is Area 13 which is the part of the limbic brain that establishes adult emotional reactions such as aggression.

Gerald Crabtree thinks we reached a peak in our intelligence about 7,000 years ago, but he doesn’t think our decline relates to our domesticating of ourselves. His argument is based on the idea that for more than 99% of human evolutionary history, we lived as hunter-gatherer article-crabtree-1112communities, surviving on our wits. However, since the invention of agriculture and cities and technology, natural selecting on our intellect has effectively stopped. He suggests that this has allowed mutations to accumulate in the genes involved in intelligence on average 25 - 65 per generation. He predicts the 5,000 new mutations in the last 120 generations, which is about 3,000 years, are the cause of our decline. He gives an interesting illustration: “I would wager that if an average citizen from Athens of 1000 BC were to appear suddenly among us, he or she would be among the brightest and most intellectually alive of our colleagues and companions, with a good memory, a broad range of ideas, and a clear-sighted view of important issues. Furthermore, I would guess that he or she would be among the most emotionally stable of our friends and colleagues. I would also make this wager for the ancient inhabitants of Africa, Asia, India or the Americas, of perhaps 2000–6000 years ago. The basis for my wager comes from new developments in genetics, anthropology, and neurobiology that make a clear prediction that our intellectual and emotional abilities are genetically surprisingly fragile.”

In today’s world we compensate for having a smaller brain or a more mutated one by using computers and technology. We also have an education system that provides the supportive environment to allow what brain power that remains to attain its potential. Education allows those strengths to be rapidly distributed to all members of our society. What we need to do is educate Athenians.

For a mash up of the video from Coursera; Genes and the Human Condition, University of Maryland by Professor Raymond St. Leger -- see below:


There are about 50 neurotransmitters, the most important ones studied are: serotonin, dopamine, oxytocin and vasopressin. Studies have shown that we are social creatures -- isolation and loneliness are toxic to our wellbeing.

This and the previous blog are notes from the Coursera: ’Genes and the Human Condition’, University of Maryland, lectures on “My Genes Made Me Do It”. What follows is from lecture 4.


One of the important neural circuits in our brain is the amygdala -- cingulate cortex neural circuit. In the diagram above, the amygdala (in red) is the threat detector and the cingulate cortex (in yellow) is connected to one of the self-awareness centers. Serotonin is used as the neurotransmitter in this circuit.

How this circuit works is illustrated in the video below. Also neurotransmitter activity during orgasm and addiction are explained as well as MAO and the 'warrior gene'. In the summary, an overall view of our genes and the environment are put into perspective.



This and the previous blog are notes from the Coursera: ’Genes and the Human Condition’, University of Maryland, lectures on “My Genes Made Me Do It”. What follows is from lecture 3.

“Psychologists don't really do consensus, but in the case of personality traits, it's hard to avoid: Openness to experience, Conscientiousness, Extroversion, Agreeableness, and Neuroticism (OCEAN for short) that constitute the sum of human personality. Each of us has our own unique coordinates depending on precisely where we fall along each of these five dimensions. Where we fall in each dimension is about 50% genetic and 50% environmental. So what do you think is the major environmental influence on personality? If you said parents then you would be wrong. Some major studies suggested that parents make only a few percent of differences in personalities and behaviors.  And the effects of family largely disappear as people get older. Criminal parents are most likely to produce criminal children. Yes, but not if they adopt the children. Likewise the children of divorced parents are more likely to get divorced. Yes, but only if they are biological children. So basically, these studies suggest that parents are overrated as shapers of values. Sandra Scarr suggests that people pick the environment to suit their characters.

You adopt the mannerisms of your peers. In the western world, at least, peers may be a lot more important than parenting. There are evolutionary reasons for this. Your peers will be your lovers, your allies, your rivals. In the long run, they're the ones who matter most. They matter more than your parents.”


It is interesting to compare Hillary Clinton’s politically slanted book: “It Takes a Village” with that of  Senator Rick Santorum’s conservative: “It Takes a Family”. Both books ignore the ability of children to make their own choices.

In a new field of study called genopolitics, it has become accepted that your political views may have a genetic component. Neuroscientists have shown that liberals and conservatives have different patterns of brain activity. In particular, there are differences in their amygdala, the part of the brain that makes emotional responses. Research has shown that people’s whose basic emotional responses to threats are more pronounced, develop ring wing opinions. Twin studies suggest that opinions on a long list of issues from religion, to gay marriage, to party affiliations have a substantial genetic component.

In the upcoming American presidential elections, it will be interesting to see how the idea that genes will influence the political outcome plays out.

Below is a mashup of the Coursera lecture:



This and the next blog are notes from the Coursera: ’Genes and the Human Condition’, University of Maryland, lectures on “My Genes Made Me Do It”.

The age old debate of nature vs nurture takes on new meaning in the light of modern genetics and neuroscience. Society’s misunderstanding of some of the implications genetics goes from one extreme where in 2009 a murderer in Italy got a reduced sentence because he had genes associated with criminality while in the US an argument was made for a higher sentence based on the prosecution’s evidence that people with particular genes cannot be cured. The gene in question was the SRY gene carried on the Y chromosome that determines maleness. There is NO GENE for criminality. Apart from the rare exceptions such as the gene for sickle cell anaemia which protects against malaria and Huntington's disease, a neurodegenerative disorder, all other traits involve hundreds of genes and the interaction with the environment (the nurture bit).

Most of the studies on the effect of the environment comes from the work with identical and fraternal twins. If one minimises the influence of the environment then the remaining differences become genetic. The paradox is that the more equal we make society, the more important we will make the genes.

Professor Raymond St. Leger presented such an example using IQ:

Present studies indicate that the heritability of intelligence, judged largely by IQ scores goes up linearly across lifespan. So from 30% in very young children, to 40%, 50%, 60%, some people even say it becomes 80% heritable by the time you're middle aged. Well that's saying that 80% of the reason that we're all different in IQ is genetic and so it suggests that genes play a majority role in IQ scores. But environment is important, particularly in young people. Remember that an average IQ is 100. So potentially the 30% of variation in IQ due to environment could be fairly significant in determining if someone has an IQ of 120, or is in the sub-normal range. However, by the time an adopted child is 18, their IQs correlate with their biological parents, and not their adopted parents. This makes the point that the environment, all that mass coaching and tiger mothering can maybe have an effect on the kid's IQ when he's young, bump him up a few notches. But as he gets older, his IQ will become ever more closely correlated with that of his blood relatives.”


The analysis of human traits has moved from using SNP’s (single nucleotide polymorphism)— a kind of barcode to the sequencing of the whole genome. This allows for genome-wide association studies (GWAS). In which tens of thousands of people are analysed for genetic variance — this also uses the non-coding parts of the DNA as well.

What are some of the highlights so far?

It turns out that some of the common conditions such as asthma and diabetes are controlled by tens or even hundreds of genes. The good news is that the pathways by which some diseases start have been identified and some entirely unexpected new pathways have been discovered. For instance a high risk version of the FTO gene has been associated with obesity and the high production of the hormone called ghrelin which makes people hungry. The study of the DNA of centenarians (people called the wellderly) have identified 5-6 biochemical pathways that are often revved up. This includes gene variants of the insulin IGF-1 pathway. As already mentioned variance of the SRY gene has been linked to violent crime.

From the twin studies a risk loci has been identified that is shared by the five major psychiatric disorders: schizophrenia, bipolar disorder, autism, major depression and ADHD. Two of these loci involve genes that are part of the calcium channels — which are used when neutrons send signals in the brain.

For a mash-up of the first set of lectures:


The symposium 'Future of Research' was held in Boston in October of this year and once again Paula Stephan recommended that biomedical graduate departments partake in birth control. "We are definitely producing many more PhD's than there is a demand for in research positions." she said.

“The biomedical research system is structured around a large workforce of graduate students and postdocs,” says Michael Teitelbaum, a labour economist at Harvard Law School in Cambridge, Massachusetts. “  Faculty members rely on cheap PhD students and postdocs because they are trying to get the most science out of stretched grants. Universities, in turn, know that PhD students help faculty members to produce the world-class research on which their reputations rest.

There are cost advantages in having a Postdoc scholar: the low salary and long hours mean that they are less expensive by almost half than graduate students of scientific staff. The hourly rate for postdocs is about US$ 16/hr compared to post-graduate students (US$ 34 - US$ 21/hr) and staff scientists (US$ 30/hr).  In the US, the NIH salary rate for a postdoc was US$ 42,840 in 2015. The other advantages of a postdoc are they have a higher skill than a graduate student and are usually more motivated than a staff scientist.

"Although principal investigators acknowledge the difficulty of securing an academic position, the system worked for them and so it is tempting to tell students that they can do it too — just another experiment, another publication or another year, and you’ll get there." says Julie Gould in the Nature article.

As for my own 'future of research' perspective, in the last 10 years of my academic career I didn't take on any more PhD students. This was partly due to the lack of resources (finances) to train a world-class scientist and also partly to my doing my bit for PhD birth control.


There is a saying that goes something like: “It is not that our plans had failed but that we had failed to plan.” This applies to the future of work. We seem to be driven down by technology as if we were in the early days of the agricultural or industrial revolution.

There are countless  conferences,  reports,  scenarios - almost every organisation feels compelled to ask:  What is the future of work?  I suggest the report by PricewaterhousCoopers is a good place to start. It offers three different lenses to view how the future might look — the Orange world where small is beautiful, the Blue world where corporate is king and the Green world where  companies care. Click here for the report.

Using constantly improving technology will mean growth but this growth will not mean more jobs — millions will be rendered obsolete. Those that can offer professional services in the digital marketplaces will secure a greater volume of assignments, market themselves, use secure payment systems and vet potential clients. One example is Upwork which connects freelancers with assignments in areas such as software programming, graphic design, marketing and mobile development. It connects some 3.6 million client businesses with more than nine million freelancers from 180 countries. This is a preview of the Orange world where large companies breakdown to a network of smaller collaborative organisations.

In my view:  in the future there will be fewer jobs and many people will not have regular jobs.  Society will need to change so that the increase in productivity of technology benefits us all — as it did in agriculture and modern factories. Instead of the profits of increased productivity going to the <1% at the top, everyone should benefit in the form of an universal living wage. The attitude and value to work would need to change -- work being a privilege, an opportunity to serve the community — a vocation.

For example contrast the attitude to work in Norway and Saudi Arabia — both living from oil.

For the full interview of Tim O'Reilly by Michael Kransy

Jennifer Doudna - co-inventor of the CRISPR technology talks about the need for 'ethics of CRISPR'.

Following on the issues raised in a previous post, Jennifer and numerous colleagues have called for an international meeting to discuss the safe use of CRISPR and the ethics of being able to create "engineer humans" as well as other genetically modified organisms (GMO's)

It is important that all stake holders (excuse the jargon), which includes you and me, understand the potential and risks of this technology and to have an informed discussion in accordance with the principles of a pluralist democracy. For more , click here.

For an elegant explanation of the CRISPR-Cas9 system watch the video.