Epidemics

Supplementary code for the paper Milton et al. - Spatial Analysis Made Easy with Linear Regression and Kernels (2019).

Excel data sheets for grey squirrel trapping field data and diagnostic viral testing results.

Data set made available contains data on the reporting delays for the submitted manuscript "Assessing reporting delays and the effective reproduction number: The 2018-19 Ebola epidemic in DRC, May 2018-January 2019". Data set contains 7 variables and 661 observations in total. Data has been extracted from the epidemic curves of the WHO situation reports and disease outbreak news using an online software, WebPlotDigitizer.

This is a set of contact-based adjacency matrix arrays (70 x 70) created from locational data of 70 feed lot cattle over 21 days. Data were processed according to 4 parameters, including Spatial Threshold(0.1665m-0.999m), temporal sampling window(called TempAgg here) (10-180 sec), and minimum sampling duration (1-4). Half the dataset includes contacts aggregated over days(array length 21), and half the dataset includes contacts aggregated over hours(array length 504).

This dataset contained the data that was used in the analysis of the manuscript entitled "Identifying human encounters that shape the transmission of Streptococcus pneumoniae and other acute respiratory infections". The study was conducted in Sheema North Sub-District (South-West Uganda) between January and March 2014. Sixty clusters were randomly selected from the 215 villages and two small district towns (Kabwohe and Itendero) in the study area, proportionally to the population size of each village and town. In each cluster, from 29 or 30 individuals randomly sampled from different households for inclusion in a nasopharyngeal carriage study, a subset of 11 or 12 individuals were selected to answer questions about their social contacts and their history of respiratory illness in the last two weeks, in addition to having a nasopharyngeal swab taken. For the social contact questionnaire, participants were first asked to list all the individuals with whom they had a two-way conversational contact lasting for ≥5 minutes during a period of approximately 24 hours prior to the survey day (from wake up the previous day until wake up on the survey day). Such encounters were defined as ‘ordinary contacts’. For each reported ordinary contact, participants (or their parent/guardian) were asked to estimate the contact’s age (or estimated age), how long the encounter lasted for and whether it involved skin-to-skin touch or utensils passed from mouth to mouth (either of those defining ‘physical contacts’). For very short social encounters (<5 minutes), which were defined as ‘casual contacts’ (e.g. seeing someone on the way, encounter in a shop etc.), participants were asked to estimate the number of encounters based on pre-defined categories (<10 contacts, 10-19 contacts, 20-29 contacts, ≥30 contacts), but not to provide further details about each contact. Next, participants were asked about respiratory symptoms experienced in the two weeks prior to the survey, including any of the following: cough, runny nose, sneezing, sore throat, difficulty breathing. Finally, after the interview was completed, a nasopharyngeal swab was taken from each participant. The data dictionary for this dataset can be found in the corresponding manuscript files

SNP alignments of M. tuberculosis outbreak data sets.

Data and Matlab scripts to replicate our experimental results

Results (percent human infection reduction) from all simulations of all scenarios presented in the research article. Results from each simulation are in a different tab.

The main code is "surv.m," which references the other 2 files provided.

Epidemic simulation in metapopulations with behavioral responses Source code of the simulations used in the work: "Modeling the effects of social distancing on the large-scale spreading of diseases". Also available on GitHub: https://github.com/paulocv/metapop-behavioral. Paper (Open Access): https://doi.org/10.1016/j.epidem.2022.100544. Preprint: https://arxiv.org/abs/2105.09697. Python environment setup with Conda We used Conda to setup Python environments for running the simulation and data analysis. A snapshot of the environment packages (with exact version numbers and builds) can be found at the file metapop_nx1_snapshot_2021-12-03.yml. It was created on a Ubuntu 20.04 OS. Important: the simulations (on sim_modules folder) are implemented using version 1.9.1 of the Networkx library, as it presents considerably higher performance than Networkx 2.x. The 1.x and 2.x libraries are not mutually compatible, therefore you should also use version 1.x or update the code to be compatible with 2.x. Alternatively, we keep a Networkx 2.x implementation on folder "nx2_version_bkp", but it is potentially outdated. For Linux OS, you can reconstruct the environment with the following command (assuming you already have Conda installed): conda env create -f metapop_nx1_snapshot_2021-12-03.yml or follow this link for instructions to create a conda env from a .yml file. For other operational systems, it is possible that you will face "package not found" due to the build numbers. You can still manually create the environment and install the packages, paying attention to their version numbers. The analysis module (nodewise_features_metapop.py), on the other hand, use NetworkX 2.x. The recommended way to work is to create a clone of the metapop_nx1 environment, activate it and then type pip install networkx==2.6.2. -------- Usage ----------- Simulations – implementation The simulations are mostly defined by two modules inside the "sim_modules" directory: metapopulation.py – Defines the Metapop class, derived from the Networkx's Graph class. The class stores all the variables of a metapopulation, and defines methods to modify it before, during and after the simulations. models.py – This is where the epidemic models are defined. A base class MetapopModel defines complete simulation routines (e.g. simulate_basic, simulate_reset_global). The rules of the epidemic model are implemented as derived classes from MetapopModel, by implementing methods such asepidemic_step_basic. The module contains two concrete epidemic models: MetapopSIR and MetapopSEIR. Simulations – running To run a simulation round with a single set of parameters, use metapop_simulate.py (see the code for usage instructions). There is an example input file on the directory "inputs". Example of simulation call (make sure to have the Python enviroment setup first, otherwise import and other errors will pop up): python metapop_simulate.py inputs/sample_input.in outputs/my_first_test/ Be aware that, if successful, this should use all the CPU threads available in your computer (up to 16). Alternativelly, you can specify the number of threads to be used as a third argument. For example, run: python metapop_simulate.py inputs/sample_input.in outputs/my_first_test/ 4 to use only 4 CPU threads. For figures 7 to 10 of our manuscript, we need to run simulations with multiple k and l values. For this purpose, you can use response_curves_script.py. Again, instructions are found in the source file. You can run it with the same example input file, for example: python response_curves_script.py inputs/sample_input.in outputs/my_second_test/ 4 This will take considerably more time, and will produce a large number of output files. Data analysis We have additional scripts to post-process our outputs from simulations. These are calc_execwise_features_parallel.py and calc_reset_statistics.py. Both use tools defined in "nodewise_features_metapop.py", which assumes that Networkx is of a 2.x version. The scripts are called without extra arguments. Please edit the inputs directly in the main() function of each script. ------- Metapopulation files -------- The code uses two files to define a metapopulation: one with ".csv" extension and another with ".edgl" (edgelist). The .csv file defines the initial number of individuals in each node. For networks in which nodes are placed in a planar space (such as the RGN used in our paper), it also contains the x and y coordinates of each node. The .edgl file contains the links and link weights of the network. Each line consists of a recyprocal link: the first two columns are the ids of the linked nodes, and the third represents the weights $T_{ij}$, as explained in our paper. You can find the syntetic RGN used in our paper at networks/sample_network.csv and networks/sample_network.edgl. ------- Toolbox ------------ The simulations and data analysis use some functions of another auxiliary library, contained in the "toolbox" folder. This is a snapshot from a personal library of helper methods in Python, developed by the same author. You can check the up-to-date library here: https://github.com/paulocv/my_tools .