Exploratory Analysis of Seized Heroin Samples Using Gas Chromatography and Chemometrics

Differences in the agricultural conditions, manufacturing procedures, addition of adulterants cause variation in the chemical profiles of a confiscated heroin sample[1]. Chemical profiling and clustering using chemometrics methods have fundamental role in the tactical and strategic intelligence about source production, and transit routes of heroin seized[2,3]. Hence, 459 seized heroin samples during 2015 and 2016 in Iran were analyzed using GC-FID and then were clustered. Our process has four steps, (a) analysing heroin samples using gas chromatography-flame ionization detector\ gas chromatography - mass spectroscopy and identifying all of volatilized component (b) determining relative peak area of each compound using docosan as the external standard (c) tabulating data set for statistical operations (d) cluster analysis. Principal component analysis (PCA) and projection pursuit (PP) as unsupervised exploratory data analysis were incorporated for clustering of aforementioned heroin samples. Projection pursuit (PP) is an unsupervised technique that seeks for an interesting low-dimensional linear projections of a high-dimensional data which are useful for clustering. Based on indications, PP scores resulted in appropriate separation of clusters/classes within the heroin data which can not be deduced from the PCA scores[4,5]. Finally, clustering of heroin samples revelas four distinct and well-separated groups which are meaningful from origin of production and transit routs point of views. Predicting the origin of seized heroin has primary concern in the tactical and technical investigation. Chemometrics analysis of a heroin data set helps in evaluation for sentencing. This study is useful for the courts concerning the possible links between samples and providing tricky information to assist in the elucidation of drug-dealing networks.