PopAffiliator 2
Prediction of an individual affiliation to a major population group based on information from a small set of autosomal STRs
To calculate the assignment of an individual to a major population group (Asia, Eurasia, sub-Saharan Africa, North Africa, Near East) the values in the form bellow should be provided. The range for the allele size was restricted to the ones published in the Short Tandem Repeat DNA Internet DataBase.
The output will indicate the probability of assignment to the major population groups. The accuracy of individual population affiliation assignment to three population groups (Asia, Eurasia, sub-Saharan Africa) is approximately 90%. The accuracy decreases to 65% when two more population groups are considered. The probabilities are computed using a machine learning model built as described in:
- Fonseca et. al., On using Machine Learning to predict the affiliation of an individual to a major population group using a forensic set of autosomal STRs. 2011. (submitted)
- Supplementary Material document is available here.
The data in ARFF format used to generate the models is (will be) available here. More information regarding the genetic profiles used is available in the previous version of popaffilator.
