Automated hypothesis generation: an AI role in science

When I was getting my PhD in Ann Arbor during the 1980’s, just staying up to date with the relevant literature to my own thesis project was a constant challenge. There was a paper magazine back then called Current Contents (CC). CC contained just that: the tables of content for all of the relevant journals (in Life Sciences). It was a critical resource because there was no other way—even then—to keep tabs on the collective scientific output.

Keeping tabs was not just for general knowledge about the field. Or even about properly giving credit to others. Rather, it was critical to the hypothesis creation. Asking the right question (at the right time) is what determines scientific success in many cases. But you can’t ask the right question without understanding whether it’s been already asked. And really you can’t ask the right question without a full understanding of what the current state of scientific knowledge is.

At the time, it was the habit, in many high impact papers, to have the last figure in the paper be a cartoon schematic that represented the author’s view of where the field was—at the moment of the paper’s acceptance into the journal. In my field of molecular neuroscience, this often was a series of shapes and arrows representing key biomolecules and pathways. It was often amusing to go from one paper to the very next that a particular group put out and see that some of the arrows would mysteriously reverse directions from the cartoon in the previous paper. This was presumably because the paper’s results along with other results had changed the thinking of the author.

In any case, that cartoon figure was always a clue into what the next hypothesis to be tested would be for a particular research group. So in a sense, you could predict the trajectory of scientific inquiry from that cartoon figure at the end of a paper.

That was the 1980’s. Our scientific knowledge base has expanded exponentially since then. One of the current versions of Current Contents is called Faculty of 1000 (F-1000). It’s on-line of course. The idea is that leaders in the field curate the papers that you should read based on your profile. It’s a great idea I guess, although science being as competitive as it is, I have doubts that the elect would give up some brilliant and undiscovered insight of a paper to the unwashed, if it really might supercharge some scientific inquiry. However, as a scientist, you have many other choices. Google Scholar comes to mind—it’s both comprehensive and I’m pretty sure it uses AI extensively to tailor its results. So machine-driven instead of human-driven (as in the case of F-1000).

However, the cartoon figure at the end of papers has become pretty obsolete (although it does still make appearances). That’s because pretty much all of science—certainly life sciences—has become incredibly complex. In my field, you can’t make a cartoon big enough to represent all the relevant biomolecules and pathways and the arrows have become incredibly intertwined because of the multiplicity of feedback loops and cross-talk links.

So not only is it difficult to glean the next hypothesis for the clever reader (even when there is a cartoon). It’s impossible for the author to do the same.

This has pushed much of science from the paradigm of Popper to exploratory research. In such science, I might read the data stream from some set of sensors, correlate that data with some other external variable (like seasonality) and publish a correlation that is intriguing. Correlation of course is not causation—we all know that.

And yet, science has the tools to do excellent hypothesis-based research. In neuroscience, optogenetics methods allow us to turn on and off neural circuits to understand their effects upon behavior. In molecular biology, CRISPR does the same for genetic circuits and networks.

The problem is not executing the research. It’s the ability to ask the right question. For biology, generating a hypothesis that is parsimonious with all of the current knowledge in a scientific discipline is challenging for human scientific superstars and downright impossible for your typical graduate student coming up with a thesis project. I believe that the same is true for any area of science where the volume of knowledge and relevant data has expanded exponentially.

But all is not lost. I think this is a perfect domain for AI as it exists today. Keeping tabs of many disparate but relevant data points and then coming up with a next move? That’s how AI’s beat humans in chess right now. So… AI in collaboration with human scientists might be a very fruitful collaboration going forward. And it may yet save hypothesis-based research.

Jasons ordered to close up shop

This is an interesting development. The Jasons Group is an elite cadre of academics who have conducted research studies for the DOD on a variety of topics over the last 60 years or so. More recently NSF has been interested in hiring the Jasons to look at the increasingly challenging climate for international collaborations between US scientists and their foreign counterparts (something that I have written a bit about). Now this news, that the Jasons contract with DOD is to be terminated. Given the views of the Administration on international collaborations of any kind, I wonder if the two things are related?

And just when you think things couldn’t get worse…

This news from today’s Washington Post on new procedures for entering the Bethesda Campus. The NIH where I did my postdoc was like the United Nations. We came from all over the globe to improve help humans stay well. In my lab alone, there were individuals from Chile, Spain, Nigeria, Italy, Israel and Australia.  Biomedical research is qualitatively different from defense R&D–Zika and Malaria do not respect political boundaries. Nor does Alzheimer’s. I hope my former colleagues in positions of authority there are listening.

Putting a chill on international science…

I saw this piece by Jeff Mervis in SCIENCE today. Basically, if you are supported by NIH and you appear to them to be more “connected” to other nation states than you have explicitly disclosed, your institution may have some explaining to do. As Jeff points out, this can be somewhat confusing, since most productive scientists (particularly in biomedical research) do their work in a manner that crosses-borders–just like Ebola or SARS. This new NIH action affects the many, not the few.  As I’ve said from my time at the bully pulpit: science is inherently international. When you publish a journal article, it is read by your colleagues all over the globe (at least if it’s good science). And that dissemination is key to producing more excellent science.

I have no problem with disclosing contacts (although there is a paperwork burden). But creating a culture of intimidation that puts a chill on international collaboration–that is a problem.

Why do we need the “driver’s license”?

When I was a trainee in neuroscience, there were plenty of examples of physicists who decided that figuring out the brain was to be their next big idea. They did just fine. You’ve probably heard of them.

Growing up with two parents who had both received their doctoral degrees in philosophy (!) but were, in actuality, running a neurophysiology lab, further inured me to the notion of being scientifically siloed by our degree title.

These days I lead two teams that aren’t doing neuroscience at all. One works on the future of AI, the other at the intersection of ecology and microbiology. At the Institute I led for sixteen years, my closest colleague was a biophysicist whose scientific passion was astrobiology. At the end of his life he was deeply involved in thinking about d-orbital catalysis in chemistry.

So increasingly, I’m coming to the conclusion that excellent science can be done by anyone with sufficient creativity and a willingness to team up with like-minded collaborators. But: with the caveat that the team as a whole has the requisite methodological skills.

This I think has implications for the future of team science–which I’ve spoken about at length.

Mid-term election: science implications I

Most of the results of the mid-term election are now in and can be reviewed on-line. Jeff Mervis at SCIENCE has a nice summer of what the changes in the House mean, here. My own sense is that with Eddie Bernice Johnson (D-TX) as the likely chair for House Science, the tenor of that Committee’s relationship with the non-biomedical US Science R&D agencies is going to improve significantly. Specifically with regards to Climate Change, and more generally with regards to a less adversarial oversight role. I think that’s probably a good thing.

NASA and probably also NSF lost a key advocate in John Culberson (R-TX) as chair of CJS, the appropriations committee responsible for the two agencies. On the other hand, NASA will probably be able to finesse the timing of when they send a probe to Europa and NSF’s contacts with Chinese science may be a bit less fettered (although the one from the White House is still pretty hawkish).

Barbara Comstock’s loss in Virginia is complex. While she could be a thorn in the side of NSF (e.g. NEON), she was extremely supportive of the DC metro area federal workforce and this benefited science agencies who depend on expert staff to keep the wheels moving.

My sense is that NIH is still coming out of this smelling like a rose. A more conservative senate may put the brakes on some hot-button research topics, but in general, I am pretty  optimistic about the biomedical sector.

 

Hana Gray

She was the President of the University of Chicago during the 1970’s and 80’s. She was also a consummate institutionalist in a way that some might think quaint in these days of social media-driven rage. Here’s a review of her recently published memoir which captures well this characteristic that is so rare among current academics. The operative metaphor is UC run as a modern-day Venetian Republic. Of course modern-day only in the sense of several decades ago and not during the Middle Ages.

Gray compares the University of Chicago’s elaborate governance structure to “the constitution of the Republic of Venice in the late medieval and early modern eras,” but praises it for “offering an invaluable means of garnering advice and discussion on all kinds of issues … with the faculty at large.” Of course, that matters only if one intends to work with one’s colleagues rather than one’s Twitter followers.