The general feature selection approach can be used in WEKA from the "Select attributes" panel.The GeneticSearch method is available there (in old versions of WEKA, like 3.6.14). You need to use it in conjunction with, e.g., CfsSubsetEval, WrapperSubsetEval, etc..
Access the package manager from the WEKA GUI Chooser with Ctrl+U (or "Tools > Package Manager"). Search for 'attributeSelectionSearchMethods'. This package contains Goldberg's genetic algorithm for attribute selection which is published here - https://dl.acm.org/citation.cfm?id=534133. It will then appear in the 'Search Method' parameter on the attribute selection window.
This will also add the source code to your weka.jar so you can edit it as you see fit for your experiment, for example you could change the fitness function to those suggested more recently, there are many discussed by Robu (2011) - https://pdfs.semanticscholar.org/7986/7b74dcb7c8982de1924e5107095e3d9b61eb.pdf
If you have the resources available, this can be used in conjuction with the Classifier Subset Evaluator in order to directly optimise the dimensionality reduction towards your specific choice of classifier.