Mosquitos + Tensors + Genetics + CS + Networks + Math + Coffee

Developed in John Marshall's Lab by:
Core Dev: Sean L. Wu, Jared Bennett
Ecology: Tomás León
Spatial Analysis: Chris De León
Data Analysis: Priscilla Zhang
Optimization: Valeri Vasquez
Former Members: Gillian Chu, Maya Shen, Yunwen Ji, Víctor Ferman, Biyonka Liang, Sarafina Smith, Sabrina Wong, Chase Violet

# Brief Description

MGDrivE is a framework designed to serve as a testbed in which gene-drive releases for mosquito-borne diseases control can be tested. It is being developed to accommodate various mosquito-specific gene drive systems within a population dynamics model that allows migration of individuals between nodes in a spatial landscape.

Currently, the model contains three modules: inheritance (A), life-history (B), and migration (D); while an epidemiological module (C) is currently under development for our upcoming v2.

# Demonstration

In this demo, we are releasing a total of 100 mosquitoes homozygous for the CRISPR/CAS9 and one with a mutation that makes the mosquito resistant to the construct. Each node in the network represents a mosquito population laid down in a spatial scenario (this could be though of as a household, house block or even city if needed). We simulate how the genetic construct would propagate across the nodes of the network if mosquitoes were slowly migrating between populations with a probability based on proximity. To watch more videos take look at our youtube playlist.

# How does it work?

The main idea behind this model is to consider the inheritance matrix of genotypes a three-dimensional structure in which each intersection point determines the ratio/probability of a specific offspring genotype (z axis) provided that a certain combination of male-female genotypes (x and y axis). This allows us to use tensors as the basis for our calculations which has many advantages, some of them being: computational speed, model's transparency and extendability.

The second novel idea in our framework is to consider the spatial layout as a network of inter-connected breeding habitats (D). By performing this abstraction we are able to transform these landscapes into distances matrices, and then into transition probabilities matrices (through the use of movement kernels). This allows our framework to be able to model arbitrary topologies in which we can simulate mosquito populations mating and migrating in realistic geographical settings.

# Installation

Our package is now available on the CRAN repository so it can be easily installed and imported with the following commands:

  
install.packages("MGDrivE")
library(MGDrivE)


To facilitate the data analysis of the results produced by MGDrivE, we provide a python package MoNeT_MGDrivE installable through pip:

  
pip install MoNeT-MGDrivE



This package provides all the functions we have used to perform the analyses on spatio-temporal gene-drives spread, and is part of our companion project: MoNeT, so have a look at the project's website for information on some of the work we are doing!