Tuesday, June 25, 2013

Y-Haplogroup Predictor

There are many haplogroup predictors but why this one? This Y-Haplogroup Predictor is unique in the sense it uses artificial intelligence for prediction. The neural network it uses is based on Back Propagation Network with Momentum having a normalized input of 48 bits, a hidden layer of 72 neurons and a normalized output of 5 bits. It had been trained to a level of net error 0.01203 using Neuroph.

Prerequisites: Java 7

Usage:
If you have JRE 7.0 installed and available on path, you can just double click on it. Then, just enter the 12 Marker STR values and click Predict Haplogroup.
You can also run the Jar file as follows:
C:\Downloads> java -jar yhaplo.jar


Download : yhaplo-predictor.jar (240 Kb)

Source Code at GitHub.

Change Log for ver 1.0:
  • Supports all major haplogroups. Subclades not yet supported.