Also, some engineering knowledge is helpful in understanding what happens here.
In a literal sense, the pressure drop (or "suction") in a tornado is insignificant compared to the wind itself just flat out blowing that hard. In an EF5 tornado the lifting force on the roof of a house, taking into account both suction and aerodynamic lift, is about 90 pounds of force per square foot - which is a lot, actually. However, the force on the walls from the wind can exceed 750 pounds per square foot. In a weaker tornado, the pressure plays an even weaker role but ALL the forces are weaker then (duh) as well. Of course, 90 pounds per square foot is more than enough to lift off your roof and have it land in the next county over.
The way a house fails is pretty straightforward, although there are a lot of variables.
Take a look at this:
http://www.spc.noaa.gov/efscale/2.html (for your typical home specifically)and this:
http://www.spc.noaa....o/ef-scale.html
If you're not aware, a supercell thunderstorm is a thunderstorm with a long-lived, persistently rotating updraft called a mesocyclone. They can actually be embedded in other storm types, such as squall lines or hurricanes, but are most often isolated and under otherwise clear air. If they get very intense, they can induce rotation in the atmosphere around them, causing the entire storm to rotate and possibly forming a mesoscale convective system (don't ask). The direction of rotation is usually anti-clockwise (cyclonic) in the northern hemisphere. This is actually NOT due to the Coriolis effect, but instead due to the overall wind patterns that form these storms.
