Code for : An evolutionary model to predict the frequency of antibiotic resistance under seasonal antibiotic use, and an application to Streptococcus pneumoniae

François Blanquart, Sonja Lehtinen, Christophe Fraser
May 2017

Description of the data and code used:

likelihood_Fourier.R: main code for data analysis.
		- For each population (B = Bedouin, J = Jewish), loads the dataset of resistance and consumption (all.B.csv or all.J.csv), approximate the prescription time series as a Fourier series
		- For each antibiotic prescribed (amoxicillin, cephalosporin, azithromycin), optimise the full likelihood, then the approximate likelihood, then compute confidence intervals using MCMC
		- save everything in .RData files

functions_fourier.R: auxiliary functions loaded in likelihood_Fourier.R
		- run_MCMC							function to compute MCMC confidence intervals
		- get.fourier 						function to obtain Fourier series coefficient of the time series of antibiotic prescription
		- get.cons 							function to obtain antibiotic prescription at any time from the fourier coefficients
		- get.integral 						function to compute the integral (the sum in equation 7 of the paper)
		- get.pred.multi2 					function to obtain predicted frequency of resistance under the full model (equation 7 of the paper)
		- get.pred.get.pred.multi.approx2 	function to obtain predicted frequency of resistance under the approximate model (equation 5 of the paper)
		- lik.multi2 						full likelihood function
		- lik.multi.approx2 				approximate likelihood function

all.B.csv: data for Bedouin childen, a csv file with 10 columns, obtained from figures 1 and 2 in Dagan et al. JID 2008
		year.cat 		the year category, from 1999 to 2004 every month (60 lines)
		res.pen 		percent strain resistant to penicillin
		res.ery 		percent strain resistant to erythromycin
		res.multi 		percent strain multiply resistant
		cons.amo	 	prescription of amoxicillin per 1000 children per month
		cons.amocla 	prescription of amoxicillin + clavulanate per 1000 children per month
		cons.amotot 	total prescription amoxicillin per 1000 children per month
		cons.ceph 		prescription of cephalosporin per 1000 children per month
		cons.azi 		prescription of azithromycin per 1000 children per month
		cons.tot 		total prescription per 1000 children per month

all.J.csv: data for Jewish childen, a csv file with 10 columns, obtained from figures 1 and 2 in Dagan et al. JID 2008		
		year.cat 		the year category, from 1999 to 2004 every month (60 lines)
		res.pen 		percent strain resistant to penicillin
		res.ery 		percent strain resistant to erythromycin
		res.multi 		percent strain multiply resistant
		cons.amo	 	prescription of amoxicillin per 1000 children per month
		cons.amocla 	prescription of amoxicillin + clavulanate per 1000 children per month
		cons.amotot 	total prescription amoxicillin per 1000 children per month
		cons.ceph 		prescription of cephalosporin per 1000 children per month
		cons.azi 		prescription of azithromycin per 1000 children per month
		cons.tot 		total prescription per 1000 children per month