Effective games from spatial structure

For the last week, I’ve been at the Institute Mittag-Leffler of the Royal Swedish Academy of Sciences for their program on mathematical biology. The institute is a series of apartments and a grand mathematical library located in the suburbs of Stockholm. And the program is a mostly unstructured atmosphere — with only about 4 hours of seminars over the whole week — aimed to bring like-minded researchers together. It has been a great opportunity to reconnect with old colleagues and meet some new ones.

During my time here, I’ve been thinking a lot about effective games and the effects of spatial structure. Discussions with Philip Gerlee were particularly helpful to reinvigorate my interest in this. As part of my reflection, I revisited the Ohtsuki-Nowak (2006) transform and wanted to use this post to share a cute observation about how space can create an effective game where there is no reductive game.

Suppose you were using our recent game assay to measure an effective game, and you got the above left graph for the fitness functions of your two types. On the x-axis, you have seeding proportion of type C and on the y-axis you have fitness. In cyan you have the measured fitness function for type C and in magenta, you have the fitness function for type D. The particular fitnesses scale of the y-axis is not super important, not even the x-intercept — I’ve chosen them purely for convenience. The only important aspect is that the cyan and magenta lines are parallel, with a positive slope, and the magenta above the cyan.

This is not a crazy result to get, compare it to the fitness functions for the Alectinib + CAF condition measured in Kaznatcheev et al. (2018) which is shown at right. There, cyan is parental and magenta is resistant. The two lines of best fit aren’t parallel, but they aren’t that far off.

How would you interpret this sort of graph? Is there a game-like interaction happening there?

Of course, this is a trick question that I give away by the title and set-up. The answer will depend on if you’re asking about effective or reductive games, and what you know about the population structure. And this is the cute observation that I want to highlight.

Software monocultures, imperialism, and weapons of math destruction

This past Friday, Facebook reported that they suffered a security breach that affected at least 50 million users. ‘Security breach’ is a bit of newspeak that is meant to hint at active malice and attribute fault outside the company. But as far as I understand it — and I am no expert on this — it was just a series of three bugs in Facebook’s “View As” feature that together allowed people to get the access tokens of whoever they searched for. This is, of course, bad for your Facebook account. The part of this story that really fascinated me, however, is how this affected other sites. Because that access token would let somebody access not only your Facebook account but also any other website where you use Facebook’s Single Sign On feature.

This means that a bug that some engineers missed at Facebook compromised the security of users on completely unrelated sites like, say, StackExchange (SE) or Disqus — or any site that you can log into using your Facebook account.

A case of software monoculture — a nice metaphor I was introduced to by Jonathan Zittrain.

This could easily have knock-on effects for security. For example, I am one of the moderators for the Theoretical Computer Science SE and also the Psychology and Neuroscience SE. Due to this, I have the potential to access certain non-public information of SE users like their IP addresses and hidden contact details. I can also send communications that look much more official, along-side expected abilities like bans, suspensions, etc. Obviously, part of my responsibility as a moderator is to only use these abilities for proper reasons. But if I had used Facebook — disclosure: I don’t use Facebook — for my SE login then a potential hacker could get access to these abilities and then attempt phishing or other attacks even on SE users that don’t use Facebook.

In other words, the people in charge of security at SE have to worry not only about their own code but also Facebook (and Google, Yahoo!, and other OpenIDs).

Of course, Facebook is not necessarily the worst case of software monoculture or knock-on effects that security experts have to worry about. Exploits in operating systems, browsers, serves, and standard software packages (especially security ones) can be even more devastating to the software ecology.

And exploits of aspects of social media other that login can have more subtle effects than security.

The underlying issue is a lack of diversity in tools and platforms. A case of having all our eggs in one basket. Of minimizing individual risk — by using the best available or most convenient system — at the cost of increasing systemic risk — because everyone else uses the same system.

We see the same issues in human projects outside of software. Compare this to the explanations of the 2008 financial crises that focused on individual vs systemic risk.

But my favourite example is the banana.

In this post, I’ll to sketch the analogy between software monoculture and agricultural monoculture. In particular, I want to focus on a common element between the two domains: the scale of imperial corporations. It is this scale that turns mathematical models into weapons of math destructions. Finally, I’ll close with some questions on if this analogy can be turned into tool transfer: can ecology and evolution help us understand and manage software monoculture?

Evolutionary dynamics of acid and VEGF production in tumours

Today was my presentation day at ECMTB/SMB 2016. I spoke in David Basanta’s mini-symposium on the games that cancer cells play and postered during the poster session. The mini-symposium started with a brief intro from David, and had 25 minute talks from Jacob Scott, myself, Alexander Anderson, and John Nagy. David, Jake, Sandy, and John are some of the top mathematical oncologists and really drew a crowd, so I felt privileged at the opportunity to address that crowd. It was also just fun to see lots of familiar faces in the same place.

A crowded room by the end of Sandy’s presentation.

My talk was focused on two projects. The first part was the advertised “Evolutionary dynamics of acid and VEGF production in tumours” that I’ve been working on with Robert Vander Velde, Jake, and David. The second part — and my poster later in the day — was the additional “(+ measuring games in non-small cell lung cancer)” based on work with Jeffrey Peacock, Andriy Marusyk, and Jake. You can download my slides here (also the poster), but they are probably hard to make sense of without a presentation. I had intended to have a preprint out on this prior to today, but it will follow next week instead. Since there are already many blog posts about the double goods project on TheEGG, in this post I will organize them into a single annotated linkdex.

Hamiltonian systems and closed orbits in replicator dynamics of cancer

Last month, I classified the possible dynamic regimes of our model of acidity and vasculature as linear goods in cancer. In one of those dynamic regimes, there is an internal fixed point and I claimed closed orbits around that point. However, I did not justify or illustrate this claim. In this post, I will sketch how to prove that those orbits are indeed closed, and show some examples. In the process, we’ll see how to transform our replicator dynamics into a Hamiltonian system and use standard tricks from classical mechanics to our advantage. As before, my tricks will draw heavily from Hauert et al. (2002) analysis of the optional public good game. Studying this classic paper closely is useful for us because of an analogy that Robert Vander Velde found between the linear version of our double goods model for the Warburg effect and the optional public good game.

The post will mostly be about the mathematics. However, at the end, I will consider an example of how these sort of cyclic dynamics can matter for treatment. In particular, I will consider what happens if we target aerobic glycolysis with a drug like lonidamine but stop the treatment too early.

Acidity and vascularization as linear goods in cancer

Last month, Robert Vander Velde discussed a striking similarity between the linear version of our model of two anti-correlated goods and the Hauert et al. (2002) optional public good game. Robert didn’t get a chance to go into the detailed math behind the scenes, so I wanted to do that today. The derivations here will be in the context of mathematical oncology, but will follow the earlier ecological work closely. There is only a small (and generally inconsequential) difference in the mathematics of the double anti-correlated goods and the optional public goods games. Keep your eye out for it, dear reader, and mention it in the comments if you catch it.^[1]

In this post, I will remind you of the double goods game for acidity and vascularization, show you how to simplify the resulting fitness functions in the linear case — without using the approximations of the general case — and then classify the possible dynamics. From the classification of dynamics, I will speculate on how to treat the game to take us from one regime to another. In particular, we will see the importance of treating anemia, that buffer therapy can be effective, and not so much for bevacizumab.

Cancer metabolism and voluntary public goods games

When I first came to Tampa to do my Masters^[1], my focus turned to explanations of the Warburg effect — especially a recent paper by Archetti (2014) — and the acid-mediated tumor invasion hypothesis (Gatenby, 1995; Basanta et al., 2008). In the course of our discussions about Archetti (2013,2014), Artem proposed the idea of combining two public goods, such as acid and growth factors. In an earlier post, Artem described the model that came out of these discussions. This model uses two “anti-correlated” public goods in tumors: oxygen (from vasculature) and acid (from glycolytic metabolism).

The dynamics of our model has some interesting properties such as an internal equilibrium and (as we showed later) cycles. When I saw these cycles I started to think about “games” with similar dynamics to see if they held any insights. One such model was Hauert et al.’s (2002) voluntary public goods game.^[2] As I looked closer at our model and their model I realized that the properties and logic of these two models are much more similar than we initially thought. In this post, I will briefly explain Hauert et al.’s (2002) model and then discuss its potential application to cancer, and to our model.

Don’t treat the player, treat the game: buffer therapy and bevacizumab

No matter how much I like modeling for the sake of modeling, or science for the sake of science, working in a hospital adds some constraints. At some point people look over at you measuring games in the Petri dish and ask “why are you doing this?” They expect an answer that involves something that benefits patients. That might mean prevention, early detection, or minimizing side-effects. But in most cases it means treatment: how does your work help us treat cancer? Here, I think, evolutionary game theory — and the Darwinian view of cancer more generally — offers a useful insight in the titular slogan: don’t treat the player, treat the game.

One of the most salient negative features of cancer is the tumour — the abnormal mass of cancer cells. It seems natural to concentrate on getting rid of these cells, or at least reducing their numbers. This is why tumour volume has become a popular surrogate endpoint for clinical trials. This is treating the player. Instead, evolutionary medicine would ask us to find the conditions that caused the system to evolve towards the state of having a large tumour and how we can change those conditions. Evolutionary therapy aims to change the environmental pressures on the tumour, such that the cancerous phenotypes are no longer favoured and are driven to extinction (or kept in check) by Darwinian forces. The goal is to change the game so that cancer proves to be a non-viable strategy.^[1]

In this post I want to look at the pairwise game version of my joint work with Robert Vander Velde, David Basanta, and Jacob Scott on the Warburg effect (Warburg, 1956; Gatenby & Gillies, 2004) and acid-mediated tumour invasion (Gatenby, 1995; Gatenby & Gawlinski, 2003). Since in this work we are concerned with the effects of acidity and vascularization on cancer dynamics, I will concentrate on interventions that affect acidity (buffer therapy; for early empirical work, see Robey et al., 2009) or vascularization (angiogenesis inhibitor therapy like bevacizumab).

My goal isn’t to say something new about these therapies, but to use them as illustrations for the importance of changing between qualitatively different dynamic regimes. In particular, I will be dealing with the oncological equivalent of a spherical cow in frictionless vacuum. I have tried to add some caveats in the footnotes, but these could be multiplied indefinitely without reaching an acceptably complete picture.

From linear to nonlinear payoffs in the double public goods game

If you recall, dear reader, around this time last year, Robert Vander Velde, David Basanta, Jacob Scott and I got excited about the Archetti (2013,2014) approach to modeling non-linear public goods in cancer. We’ve been working on this intermittently for the last year, but aim to focus now that I have settled in here at Moffitt. This means there will be a lot more cancer posts as I resume thinking careful about mathematical oncology. Although I didn’t update the blog in the summer, it doesn’t mean that nothing was written. The work below is mostly from when I visited Tampa in late July. As are these two blackboards:

In this project, we are combining growth factor production (Archetti, 2013) and acidity (2014) as a pair of anti-correlated public goods. The resulting dynamics cannot be understood by studying just one or the other good. The goal is to explore the richer behaviors that are possible with coupled social dilemmas. At the start of the year — in my first analysis of the double public goods game — as a sanity check I considered the linear public goods and . After a long meeting with Robert a few month ago, I think that these were misleading payoffs to consider. I jotted these notes after the meeting, but forgot to release them on the blog. Instead, you get to enjoy them now while I refresh my memory.

Pairwise games as a special case of public goods

Usually, when we are looking at public goods games, we consider an agent interacting with a group of n other agents. In our minds, we often imagine n to be large, or sometimes even take the limit as n goes to infinity. However, this isn’t the only limit that we should consider when we are grooming our intuition. It is also useful to scale to pairwise games by setting n = 1. In the case of a non-linear public good game with constant cost, this results in a game given by two parameters — the difference in the benefit of the public good from having 1 instead of 0 and 2 instead of 1 contributor in the group, respectively; measured in multiples of the cost c. In that case, if we want to recreate any two-strategy pairwise cooperate-defect game with the canonical payoff matrix then just set and . Alternatively, if you want a free public good (c = 0) then use and . I’ll leave verifying the arithmetic as an exercise for you, dear reader.

In this post, I want to use this sort of n = 1 limit to build a little bit more intuition for the double public good games that I built recently with Robert Vander Velde, David Basanta, and Jacob Scott to think about acid-mediated tumor invasion. In the process, we will get to play with some simplexes to classify the nine qualitatively distinct dynamics of this limit and write another page in my open science notebook.

Double public goods games and acid-mediated tumor invasion

Although I’ve spent more time thinking about pairwise games, I’ve recently expanded my horizons to more serious considerations of public-goods games. They crop up frequently when we are modeling agents at the cellular level, since interacts are often indirect through production of some sort of common extra-cellular signal. Unlike the trivial to characterize two strategy pairwise games, two strategy public-goods have a more sophisticated range of possible dynamics. However, through a nice trick using the properties of Bernstein polynomials, Archetti (2013,2014) and Peña et al. (2014a) have greatly increased our understanding of the public good, and I will be borrowing heavily from the toolbag and extending it slightly in this post. I will discuss the obvious continuation of this work by considering more than two strategies and several public goods together. Unfortunately, the use of public goods games here — and of evolutionary game theory (EGT) more generally — is not without controversy. This extension is not meant to address the controversy of spatial structure (although for progress on this, see Peña et al., 2014b), but the rigorous qualitative analysis that I’ll use (mostly in a the next post on this project) will allow me to side-step much of the parameter-fitting issues.

Of course, having two public goods games is only interesting if we couple them to each other. In this case, we will have one public good from which everyone benefits, but the second good is anti-correlated in the sense that only those that don’t contribute to the first can benefit from the second. A more general analysis of all possible ways to correlate two public-goods game might be a fun future direction, but at this point it is not clear what other correlations would be useful for modeling; at least in mathematical oncology.

By the way, if you are curious what mathematical oncology research looks like, it is often just scribbles like this emailed back and forth:

I’ll use the rest of this post to guide you through the ideas behind the above sketch, and thus introduce you to the joint project that I am working on with Robert Vander Velde, David Basanta, and Jacob Scott. Treat this as a page from my open research notebook.