In the graph there is a parameter for fileList. This could be a list of file paths separated by commas or a folder.
If a folder is specified it will use every product it can find within that folder.
<parameters>
<fileList>E:\mydata\</fileList>
</parameters>
You could also put a variable ${inputFolder}
<parameters><fileList>${inputFolder}</fileList></parameters>
and then replace that variable from the command line
gpt mygraph.xml -PinputFolder=“e:\mydata”