Evolutionary biology
Sexually Antagonistic Selection
2019 project on sexually antagonistic selection in Dr. N. G. Prasad's Evolutionary Biology Lab (EBL), IISER Mohali.
Research project (unpublished) · 2019
Interactive abstract
Play with the idea
This interactive diagram needs JavaScript. The schematic and write-up below explain the same idea.
Layman view
In plain words
Draft — to be verified. Written in plain language from the paper and project notes on this site.
Males and females share almost all of their genes, yet they can benefit from different versions of the same gene. A version that helps males may harm females, and the other way round. Biologists call this sexual antagonism, a tug of war between the sexes over a single gene.
Mathematical models predict when such a tug of war can keep both versions in a population instead of letting one win. This project rebuilt published models of sexual conflict in fruit flies (Drosophila) in two independent ways: once with the equations from the papers, and once as a simulation of individual animals that is not allowed to use those equations.
Statistical tests then ask whether the simulated populations behave as the equations predict. Agreeing from two independent directions is a stronger check than either one alone.
Expert view
The work, step by step
-
01
Context
A 2019 project on sexually antagonistic selection in the Evolutionary Biology Lab (EBL) of Dr. N. G. Prasad, IISER Mohali.
-
02
Question
The repository asks two questions. Do populations of simulated organisms, built only from biological rules, produce the patterns that the published mathematics predicts? And for published sexual-conflict results from the lab, does the model predict what the experiments found?
-
03
Computational models
The public repository rebuilds published models of sexual conflict: a two-locus model for the resolution of intra-locus sexual conflict from a poster by the lab, Parsons' (1961) invasion conditions with Owen's (1953) equilibria, and a "mother's curse" cytonuclear model. Each is solved twice: as deterministic recursions, and as an individual-based simulation that never imports the analytical maths. Statistical tests then check whether the simulated populations reproduce each prediction. A second part models published sexual-antagonism results from the lab with a hybrid analytic, individual-based and emulator approach.
-
04
Results (from the repository)
From the repository's generated report and its plain-language summary of the hypothesis tests.
The deterministic reconstruction reproduces the poster's polymorphism formula, p* = ½ + ½(1/b − 1/a), and all five summary points of its stability analysis. Its invasion thresholds come within about 0.002 of the values read off the poster's bifurcation plots.
In the individual-based simulation, a rare modifier allele spreads or declines in the direction the mathematics predicts in almost every parameter cell that reached a verdict. The one disagreement lies within about 0.5% of neutrality, where drift dominates.
A caveat the deterministic model cannot show: balancing selection at the polymorphism is very weak, so in finite simulated populations drift fixes one allele within hundreds to thousands of generations unless the population is large.
For the lab's published sexual-conflict results, about 40 formal hypothesis tests found that the model's genuine predictions are right in direction and roughly in size. The evidence is modest, because the experiments are small. Where the data show more than the model contains (vial-to-vial scatter, a sex difference in offspring survival, genotype effects on courtship), the model fails, which points to biology it is missing. Many earlier "matches" were not evidence, because those tests could not have failed.
Schematic
How it works
Contribution
My role
In summer 2019 (May–July) I was a research intern in Dr. N. G. Prasad's Evolutionary Biology Lab at IISER Mohali. I developed mathematical models of gene-frequency patterns of sexually dimorphic traits in Drosophila melanogaster, simulated population-genetic evolution in MATLAB, and analysed data in R.
Publication