PhyloCov

An interactive web-based framework for exploring ecological covariates for integrated phylodynamic models

Explore
↓

Background & Aims

Quantifying the ecological drivers of pathogen transmission remains a key challenge in integrated phylodynamic modeling. PhyloCov bridges this gap by providing an accessible, web-based platform built on Google Earth Engine to host, explore, and extract time-varying ecological covariates.

The platform supports two distinct analytical pipelines to enhance the application of transmission models: continuous phylodynamic analysis (via GeoJSON extraction) and discrete phylodynamic analysis (via Shapefile input extraction for GLMs).

Pipeline 1: Continuous Phylodynamics

Explore and extract ecological covariates in GeoJSON format. This path is designed for continuous-style phylodynamic analysis, generating environmental data perfectly formatted for post-hoc and landscape analyses using tools like Seraphim.

Open in new tab ↗

Pipeline 2: Discrete Phylodynamics (GLM)

The second pipeline allows users to input a point or polygon shapefile to extract covariate data formatted specifically for discrete phylodynamic approaches using Generalized Linear Models (GLM) in software like BEAST.

GLM Data Format Integration

In this context, the terms predictor and covariate are interchangeable. PhyloCov generates predictor files as comma-separated value (CSV) matrices matching the exact format required by BEAST for setting up GLM-CTMC location traits:

  • Pairwise Predictors: Output as a square matrix (e.g., flight matrix, intra-continental distance matrix). The different locations serve as alphabetically ordered row and column names, with rows denoting the origin and columns the destination.
  • Origin-Destination Predictors: Output as a two-column table with alphabetically ordered location names mapped to predictor values.

Once extracted, these files can be easily loaded into the "Sites" tab in BEAST by clicking "Setup GLM" and "Import Predictors". Within BEAST, these predictors can be log-transformed and standardized to inform the rates of spread between geographic locations.

Research Impact & Future Updates

By streamlining ecological covariate extraction and formatting them specifically for continuous or discrete downstream analyses, PhyloCov facilitates the reproducible integration of environmental drivers into infectious disease modeling.

Future updates will focus on pathogen-specific categorisation and API functionality for large-scale data extraction across both pipelines.