Question regarding some HClust options 

I hope I could get a clarification on what --sperc & --fperc option do... What does "Percentile of feature/sample value distribution for sample selection" mean? Furthermore, if I could get an explanation on what ftop or stop does too. Using the ftop option does not seem to pull out the top abundant features in my dataset. 

Thank you, any explanation would be greatly appreciated. 
