by Rose and Losia
​
Originally published as an online "Behind the Paper" article at https://socialsciences.nature.com/users/177900-rose-o-dea/posts/39046-slow-and-steady-our-meta-analysis-on-greater-male-variability-in-academic-performance

The paper, authored by us (Rose and Losia) and our co-authors Michael and Shinichi, is freely available in Nature Communications: https://bit.ly/2N2ymie.

The seed for this paper was planted five years ago. Michael was sat at Losia and Shinichi’s kitchen table in Dunedin, New Zealand, listening to Shinichi describe a new meta-analysis technique he’d been developing. This technique could test for differences in variability between two groups, while controlling for differences in means. Michael suggested we use this technique to test the greater male variability hypothesis.
It would be easy. Losia found a recently published meta-analysis that tested for gender differences in average school grades; we’d do a re-analysis. It would be a simple project for Michael’s honours student, Rose, to learn some meta-analysis basics; she was joining Shinichi’s new lab in Sydney, Australia, at the beginning of 2015. The paper would be finished (we all thought) within a year, before Rose started a PhD. But it took that long just to get the data, never mind analysing and writing up our results.

The descriptive statistics of studies included in the original meta-analysis were not available. So, we downloaded the papers and re-extracted the necessary data, and contacted the authors of papers that hadn’t reported descriptive statistics. We also updated the original systematic search to find papers published since mid–2011. Losia and Rose each screened over 3,000 abstracts (thanks abstrackr), and by the start of 2016 we finally had our comprehensive dataset.

But by then, this was a side-project. Not part of anyone’s grant or thesis, it was often relegated to the bottom of our priority lists. Having each other as co-authors was essential: the project never fully stagnated because there was someone else to keep it moving. Had we been doing the project alone it would probably lay abandoned.
We were also challenged by our ignorance of the field – we are evolutionary biologists and behavioural ecologists, not social scientists or psychologists. It took time to find, and read, studies that informed our introduction and discussion. Often we were missing jargon that was the key to unlocking this information. Here we are thankful to popular science communicators for bridging these gaps (e.g. the term ‘occupational segregation’, which appears in our introduction, was learnt on an episode of Freakonomics Radio), and to psychologists who commented on earlier drafts.

by Dony Indiarto

Earlier this year, a team of researchers published a study on Indonesia’s Bajau people that made a highlight in evolutionary biology. Bajau people have been widely known as extraordinary divers. The study unveiled that these people have significantly bigger spleens compared to those live in the nearby farming communities. Bajau people spend most of their days freediving and hunting underwater for generations. The researchers also identified 25 distinctive genes that are plausibly associated with the sea gipsies’ unique lifestyle, as reported in Cell. No doubt, this research has revealed crucial evidence on human adaptation under extreme condition. Scientifically, there has been no flaw or concerns about the research and its findings. Nevertheless, the publication has been disconcerting for some Indonesian scientists, who revealed concerns over western scientists’ behaviours.

Helicopter science
The story above reflects a bigger social phenomenon, the so-called ‘helicopter science’, where foreign scientists come and collect data for a short period of time in a particular local area. They then go back home, analyse and write their research in their institution, often disconnected from the local contexts, from the local knowledge, and from those who have been focusing their research in longer term, often for their entire career. Two potential misses: (i) the research results might be biased due to small samples and also lack of local contexts; (ii) without true collaboration with local scientists, if any, long-term research for covering breadth and depth of exciting findings will not be feasible. The study may bring some progress to science, but by leaving behind the local researchers, the knowledge and capacity gaps among the scientific community will be widened, primarily due to lack of funding and resources to support scientific research in developing countries. The problem is magnified specifically for basic natural science that needs advanced technology and expensive lab works, such as in genetic studies.

This issue often happens in biodiversity-rich developing countries, where data are scarce and badly needed, but funding some particular research topics are not yet a priority. These countries are often alleged for enforcing complicated procedures and unclear regulation in facilitating and hosting foreign researchers and in permitting samples to be taken out of the country. While this is factual, it does not tell us the complete story. Cultural and language barrier can also create misunderstanding and unmet expectation between scientists. The front desk officers who implement the procedures are blind to the reason behind the regulation. Local counterparts are often employed as logistic support or leg workers and ended up having narrow space, if any, in participating in doing science. Not unlike Trump’s current policy for balanced trade, developing countries’ protection policy can retaliate back by imposing stricter regulations for research by foreign scientists. This is also similar to meta-analysis scientists who are called as “research parasites” by other scientists who collect primary data. As a result, this tit for tat will slow down new scientific discoveries in under-researched areas.
 
Call for inclusive science
“Helicopter science” is just one form of implicit biases and inequalities that do exist and persist in the scientific community. The wall between countries in doing science should be removed through collaboration driven by ethics.  The scientific partnership should be built on mutual respect and benefit. Of course, this is easier said than done. Scientists in developing countries are hindered by limited financial and institutional support, technical skills, as well as exposure to the advances in scientific discoveries. Practical or applied research is often more relevant for the locals than basic research, and therefore the collaboration in such areas will be truly fruitful.

On the other hand, local scientists should actively engage in reducing “helicopter science” practice by being pro-active. One good example is from African scientists who have taken a step forward by declaring the ethical handling of samples for genomic studies. Like everything else, the sociology of science will be rich and flourished when diversity is cherished.  In reaching so, collaborative research has to bring benefit to all parties.

Now there is no one way to give a good presentation, but below I have compiled some tips and words of wisdom I collected throughout the years on how to make presentations an enjoyable experience for yourself and your audience too.
 
Be prepared. Be really prepared

My first and foremost tip for anyone who is uncomfortable with public speaking is to plan ahead so you can have enough time to prep. I am an anxious person, I literally schedule a two-week reminder for when I have to give a talk, so I have enough time to think about the audience, the story, the format and style.
 
Hello, meet Practice and Feedback

When it comes to really nailing a talk, practice and feedback are the key. Although too much practice can also be a bad thing! You don’t want to come across as too scripted or even bored because you have practiced too much! Testing out your talk helps you getting into the flow of speaking in front of other people, it also helps you to identify parts of the talk where you are stumbling on, etc. Recording yourself is also another good way to do this.
 
Organise a practice session with your lab group, your office buddies and other students that might be presenting at the same event! Seeing other people’s talks can also spark ideas for your own. Practice with people that you feel comfortable to receive (and give) honest, constructive feedback, because that is the only way to improve. I find practicing with people that are not directly in my own field invaluable. I practice with my writing group which consists of a bacteria bioinformatician, a marine biologist, a plant geneticist, an avian evolutionary biologist and a invertebrate ecologist and honestly… this is where I get the best feedback. Explaining concepts to a broader audience helps you communicate your ideas more effectively. If you can explain a highly specialised jargon-y concept to someone not working in your field, then you are winning at science communication!

Slides are your sidekick

Depending on what type of speaker you are, the amount of text on the slide is really up to you. If you need some bullet points as prompts, then do it! But let’s all remember a time where we painfully watched someone read off their slides… again balance is important!
 
I never use animations. There is just too much room for error when transferring your files between different operating systems etc. However, I do use multiple slides to sequentially pop up elements on my slide. Like a flipbook, each slide is almost the same of the previous slide but with some small adjustments. (See here for some examples). I end up with a lot more slides but now I have fail-proof animations! Another trick is to export and transfer your slides as a .pdf so all your customised text and figures don’t move because of incompatibilities between different computers
 
Sequential ‘pop ups’ help you guide your audience’s attention to exactly what you want them to focus on. Do all the work for your audience by walking them through your argument step-by-step. Never assume they know what you mean. I use a lot of boxes to conceal parts of graphs, I cover my data, and first explain the axes, I even put some made-up data to show them how to interpret these when I finally reveal my own graph. Here’s one of my recent presentation as an example
 
Questions time is like a box of chocolate

The presentation, you can practice, you know what will happen. But with question time…it is a mixed bag. One trick, I’ve been told, is to anticipate what kind of questions you might get.
Think about: “If I watched my talk, what would I ask?” Practice sessions can help you identify these as well. My supervisor, Dan, even tries to strategically guide the audience to ask a particular question by providing less detail during the talk(!). Having an expectation of the range of questions you might be asked allows you think of a response in advance and even prepare some additional slides that you can tack on to the end of your presentation to really address the question thoroughly. Genius!
 
I wanna talk like you

One thing I find really helpful is to think back to a memorable presentation you’ve seen and identify what made that presentation so effective. I find myself thinking back to a presentation by an old lecturer A/Prof Greg Holwell. He effortlessly conversed with the audience with stunning photographs on his slides. I think of Dr. Camilla Whittington who had complete control of the whole room while she posed a pressing problem in her field and seamlessly addressed it with her own research. I am reminded of a keynote by Dr. Anne Gaskett, whose personality and casual presenting style made her research even more engaging. Drawing ideas from those that inspire you will help you develop your own style. Channel their confidence and composure!
 
Everyone responds to public speaking differently. Some thrive in the arena, while others just need “fake it before they make it”. I, for one, definitely fall in the latter category. However, over the many conferences I’ve been to I have managed to see presentations as a fun thing to do where you can show off your own work! Remember to enjoy the process and know that you are the expert of your own work. 

Conference visit and talk in Paris  (https://cee2018.sciencesconf.org/)

I landed in Paris on 18th of April (with a whole-Paris public transport strike waiting there to welcome me), so I could only make the latter part of the third day of the conference. Last two days of the conference I attended had the most exciting sessions for me and our research group. Talks were under two major categories: (a) discussions on how can technology contribute to evidence synthesis, and (b) presentations / demonstrations of software tools that assist evidence synthesis. Some key discussion points were: (i) using standard formats and tools to report systematic review process, (ii) visualisation and mapping of bibliographies, (iii) automating abstract / full-text screening using text mining, and (iv) potential future uses of machine learning in evidence synthesis.

My talk for the conference extended the fourth discussion point listed above, and my talk was titled “Deep Learning - Potential Uses in Systematic Reviews”. This was my first conference talk without presenting a paper, so it was a whole new experience for me to present something as a short-lecture format. The talk started by introducing the basics of machine learning and deep learning, then the potential uses of the introduced techniques in different stages of systematic reviews. My talk went very well, I got some overwhelming response (and also some good questions). The best comment I got was simply: “I learned from your talk!”. After a few tech talks, there was a technical discussion panel and I could be a part of that, and I enjoyed it. Machine learning was one of the hot topics during the discussions and introduction of proper database systems to keep track of evidence syntheses was equally popular.

Hackathon in Stockholm (https://evidencesynthesishackathon.com/)

After the conference, my next destination was evidence synthesis hackathon in (magnificent) Stockholm, to which I (and some other participants) carried some ideas from the Paris conference. There were hackathon participants who came from all over the word, some from extremely popular places for technology (look up the list of participants here). Some of them were hackathon enthusiasts (who try to participate in every possible hackathon), tech-gurus from industry (well known companies), computer scientists (from well known universities), and academics who mastered evidence synthesis. A computer science (>=80%) and evidence synthesis (<=20%) hybrid like me could fit in well, fortunately.

At the beginning, we all gathered in for a round-table discussion and the hackathon organisers were presenting the open problems in evidence synthesis that we ‘hackers’ could help to solve, and we all pitched in our ideas / plans as well. The whole set of participants then divided into 4 main streams according to the sub-task within evidence synthesis they were trying to ‘hack’: (i) reporting of evidence synthesis processes, (ii) improving search queries, (iii) data extraction from pdf articles and indexing the coded documents, and (iv) data visualisation using systematic maps.

Four participants, including me, were working on the third stream listed above. Our aim was to implement a software tool that assists efficient content extraction from pdf documents during data coding. The most significant barrier during this task was that the pdf files are embedded in highly abstract and protected manner. The main contribution of the prototype application we prepared was accessing contents in pdf articles selectively. The tool was developed around ReactJS JavaScript framework, therefore, it is suitable for deploying as a web application, in a local virtual web-server in a desktop environment or in a centrally hosted web-server. This application takes in a CSV file with fields to be extracted as headers, and loads pdf files from a server folder. The content coding can be performed using a right-click menu that brings up list of fields, and then saves the fields, selected contents / values from the pdf and any user comments back to the CSV file as a new row per single pdf.

Hackathon concluded with us presenting and demonstrating our implementations, concluding discussion session, and a nice dinner outing with the whole group. I mostly value the opportunity to meet and talk with people of different career and academic backgrounds, and the things I could learn from and share with them.

Acknowledgments
Shinichi for nominating me for both events, introducing me to the key people, and supporting my conference visit and the talk. Neal Haddaway at the Stockholm Environment Institute and Martin Westgate at the Australian National University (https://evidencesynthesishackathon.com/organisers/), for organising the hackathon and the technical discussion sessions at the conference, inviting me for both events and funding my visit to the hackathon

by Alex Aloy
​
David Moore’s Developing Genome is almost comparable to popular non-fiction books that I have read in the past. Conveniently and witfully written as Jared Diamond’s Guns, Germs and Steel or Tim Flannery’s best-seller about the world’s fate in the future on the brink of a changing climate in The Weathermakers. There is a clever “ladder-like” introduction of biological terminologies as you go along reading without being overly technical and scientifically jargonised.

The book exhibits a perfect balance of introducing hard facts and concepts while citing related synonymous daily life scenarios. A literal example is how Moore tried to illustrate certain sections of the DNA strand into familiar household terms such as a “yarn spool.”  I like how he put humorous connotations likening DNA processes like expression of a resulting trait or cellular product to “a party that came altogether spontaneously.” Overall, the writing style made reading the book enjoyable and interesting at least for the uninitiated with the subject topic.
 
Developmental biologists: epigenetics rings a bell
The word “epigenetics” was first introduced by Conrad Waddington in the 1940s to refer to the intrinsic interaction between genes and molecules at the cellular level, leading to phenotypic outcomes. However contemporary scientists would extend this definition as the interaction of the genetic (nuclear) material to extracellular environmental factors that are spatially situated “on top” of the genome (thus the epi word). 

The book provided a good literature review, mostly over the last decade, on the advances of epigenetic research. Most of these highlighted the role of how genes (and consequently traits) are expressed or repressed in clinical psychology and bio-medical experiments. Curiously, the process is not gene-independent but is largely influenced by its interaction with the environment. Specific examples range from serious diseases/disorders (e.g. Prader-Willi, Angelman syndromes) to fur coloration pattern in cats. In addition, a number of important biological development processes such as female human X-chromosomes inactivation have been linked to epigenetic causes. So developmental biologists knew this concept right from the start!

Developmental stage is important because environmental effects influence the expression of genes mostly in early life. In practice, if the goal is to intervene or reverse these phenotypic expressions, the key is to look and act on these at different developmental stages. This is where epigenetics comes in to elucidate the adverse effects of early life exposure to external factors such as bad (or good) experience, memory and diet deprivation during developmental stages.
 
It is indeed Nature + Nurture
Turning the book’s page one at a time has been revelational. All along, I have been versed with a misconstrued (mostly during my biology undergrad days) idea that sole genetic makeup always determines what traits, physiological or even behavioural tendencies one have in later life. Apparently, this is one of the main arguments raised by Moore about the need to dismiss popular metaphorical fallacies associated with how we use and interpret the words DNA and genes. For example, genes were popularly viewed as a “blueprint system”, a “program” (computer) and even likening it to a film script.

These in a way are misleading metaphors, because they convey a “grandmaster” plan or seem to entail a pre-deterministic fate of how genes should function and develop. Epigenetic systems are much more of a two-way process where genes’ activities are largely influenced by external factors and vice-versa. To quote, “it is not what (genes) you have, but more importantly how those genes perform and work.”

The book enumerated a wealth of supporting evidence from recent and emerging research on behavioural epigenetics that link the effect of environmental factors to genetic makeup at the individual level (i.e. DNA methylation, histone modifications). There were early experiments such as investigating consequences of epigenetic states of good or bad experience (from parental care) to eventual offspring behaviour. However, it is not clear how epigenetic factors affect individuals in succeeding generations on the population level in natural environments or the mechanisms behind it.
 
A retrospect: Darwin’s finches
The idea that our genome continuously communicates with the environment “epigenetically” puzzles me at this stage. But one thing I realise is that isn’t this supposed to be how organisms ultimately adapt and evolve under Darwinian synthesis of evolution? This reminds me of the outcomes of some classic empirical work in ecology and evolution, such as the long-term study of Darwin’s finches by the British couple Peter and Rosemary Grant in the Galapagos Islands.

Decades of field data allowed them to observe changes in feeding adaptations exhibited as adjustments in morphology and habitat use due to an ever-changing environment. What happened really was successful adaptation and transgenerational inheritance of novel phenotypic changes, which maybe are partly mediated epigenetically.
 
Re-inventing Darwinian and Lamarckian views of evolution
One of the main arguments of the book is the need to revisit the universally accepted theory of evolution or Neo-Darwinism, which is based on changing gene frequencies across a particular genome in space and time. Following the recent advances in mechanistic study of epigenetics, there is a need to explore how such processes play a role in broad evolutionary biology questions, like speciation and natural selection. To quote:

“There is a need to understand the relation between heredity and on the premise that natural selection has different results across a spectrum of heritable variations. With epigenetic systems in place and recognised, a research that will integrate development and evolution would result to a more powerful Darwinian theory.”

Amazingly, this realisation brings back again (after all) Charles Darwin and his “Origin of Species”, which is compatible and complementary to the Lamarckian view. The unlikely return to Lamarckian views of evolutionary mechanisms of inheritance, which states that it should encompass “both genetic and non-genetic developmental resources”, is unanimously sensible at this point.

In the end, Moore believes that the burgeoning field of epigenetics has finally provided the mechanistic explanation to Darwin’s theory. As epigenetics becomes more prevalent and influential in so many fields, such as in medicine, pathology and law (I am hoping so!), it is predicted that it will be a “monumental” step towards how we view life. Its impact is potentially revolutionary in the sense that it transcends from the minute DNA molecules to social and cultural hierarchies of human civilisation.

by Rose

The academic workforce is very mobile. A quick perusal of the current and former members of I-DEEL shows that our lab is no exception; there are links to several universities around the world. I do not experience wanderlust, so the frequent necessity to uproot one’s life to stay in academia is one of the least appealing aspects of this career path. Still, I believe there’s merit to the cliché ‘get out of your comfort zone’, so I recently defied most of my basic instincts by moving halfway around the world, to join the Hendry Lab at McGill University for one year of my PhD.
 
I’ve been thinking about the costs and benefits of moving institutions. On the costs side, all the logistics take up big chunks of time before, during, and after the move, which reduces productivity. Then, there are also the costs of leaving behind friends, family, and geographically-confined-interests, and the exhaustion of adjusting to a new home.
 
As for the benefits? The literature speaks of increased human and social capital, which simply means developing new knowledge and skills, and forming beneficial social connections. But I also think there might be a benefit through a motivation boost, like task switching, but on a large scale.
 
Psychologists used to think that humans had a finite amount of willpower, and if we used it up on one task (e.g., writing during the day) we would have less willpower to give to other tasks (e.g., exercising in the evening); the technical term is ‘ego depletion’. The recent surge in transparency and replication in the social sciences, however, has shown this could be false. Our willpower may be depleted if we are only doing one task, but if we switch tasks during the day, or take breaks, then we can continue to work at a productive level.
 
Moving institutions forces you to take a complete break from work during the actual moving phase, and then – because you are in a different place surrounded by different people – all your tasks feel slightly different. I have no idea whether there’s any merit to this idea, but there is some evidence linking mobility to productivity.
 
Whether or not moving is good for productivity, it is the current reality in the lives of many academics, and there are undoubtedly many personal benefits to be gained. So when I return to I-DEEL in 2018, I will hopefully bring back not only new academic skills, but also the ability to cycle through snowstorms, and an encyclopaedic knowledge of Quebec’s best cheeses.

by Gihan Samarasinghe

We are living in the age of digitisation. Each day hundreds of Terabytes of data are being dumped into digital storage networks (World Wide Web is the largest of them all). Unfortunately, most of them are unstructured and randomly ordered. Think how much of unordered, unstructured information can be shared on Twitter and Facebook within few seconds!

Discovery of the potentially important information hidden among unstructured data stored on the internet has always been an active research topic. That is where Natural Language Processing (NLP) comes to aid. NLP is one of the strongest branches of Artificial Intelligence (AI), where recordings/articles containing spoken or written human languages are analysed by computer algorithms to extract and represent useful information in a structured manner.

Role of Text Mining

There are many data mining and machine learning techniques available for the purpose of knowledge discovery in natural language contexts, and the term text mining is generally used for the techniques used to analyse text-based data sources. What basically happens in text mining is the statistical prediction of the relevance of a text-based data instance with respect to a particular subject of interest, either based on prior learned inter-relations of sets of keywords or predicted correlations among newly seen word groups. Those predictions are usually represented as lexical, syntactical and semantical analyses of text contents.

Can Text Mining Benefit from Deep learning?

The idea and evolving implementations of the family of machine learning techniques called Artificial Neural Networks (ANN) existed since the 1940s. ANN algorithms imitate and are influenced by the neural networks of animal brains. Traditional classification techniques in machine learning need a set of pre-defined features to learn from a presented instance (the process called Training). ANNs, on the other hand, contains several sequential middle layers (called hidden layers because the feature definitions are unexplained and fuzzy in these layers), that defines the features by themselves by observing the important aspects of training instances.

The rate of learning depends on two factors mainly: (i) the number of hidden layers and (ii) number of training instances. Until the early 2000s, the vital bottleneck was the unavailability and the higher cost of computational power and storage. When these issues were addressed with revolutionary hardware inventions such as fast computations using Graphical Processing Units (GPUs), faster digital read and write on Solid State Drives (SSDs), etc, ANNs also gained a rapid enhancement. This was the beginning of the popular term Deep Neural Networks, that is nothing more than ANNs with a larger number of layers (deeper) and hence a better learning rate. Day by day, various fields have benefitted from deep neural networks with the availability of millions of data and vast computational power. Meanwhile, Deep Neural Networks themselves are evolving with optimised algorithms and frameworks, therefore becoming more and closer to the way how human brain thinks and understands things (although machine learning is still far behind our brain). The following figure gives a nice overview of how Deep Learning becomes better when there is more data to train with, and capability to handle more data.

Text mining couples with deep neural networks brilliantly, due to the nature of the understanding needed in text context classification and knowledge discovery. Deep learning-based text mining techniques have become quite popular and successful in self-discovery and interpretation of interesting features (keywords) and have already shown promising signs of becoming superior to traditional machine learning techniques.

Future of Systematic Mapping

Systematic mapping of evidence for a topic of interest, in literature, is an imperative potential usage of text mining, when large networks of millions and billions of text articles are available, and yet to be summarised or collected meaningfully. In this process, defining a proper set of keywords or search terms is a tricky task, and they are heavily subjective and ambiguous. Therefore, deep neural networks are the way to go with future knowledge hubs and structured literature collections. Not only better classification of text documents using a self-defined set of keywords (with a proper training process), but also visualising and presentation of the learned interpretations will be potentially interesting outcomes of deep learning based literature surveys.

Further Reading

1. Collobert, R., Weston, J., Bottou, L., Karlen, M., Kavukcuoglu, K. and Kuksa, P., 2011. Natural language processing (almost) from scratch. Journal of Machine Learning Research, 12(Aug), pp.2493-2537.
2. Moens MF. Argumentation mining: How can a machine acquire common sense and world knowledge?. Argument & Computation. 2017 Jan 1(Preprint):1-4.
3. Hirschberg, J. and Manning, C.D., 2015. Advances in natural language processing. Science, 349(6245), pp.261-266.
4. Jason Brownlee, August 16, 2016. What is Deep Learning. machinelearningmastery.com
   
5. Andrew Ng. Deep Learning, Self-Taught Learning and Unsupervised Feature Learning. YouTube Video