Consider the following mechanism: a black hole occasionally falls into the supermassive black hole in the centre of a galaxy. This causes gravitational waves to be emitted that take e.g. 100,000 years to propagate through the galaxy (i.e. its diameter in light years). Now, even though several (up to several thousand or million is a possibility) solar masses are converted into energy during the merger, it seems clear from rough ballpark estimates that the gravitational waves distort space by something on the order of metres at the distance of the radius of a galaxy. Which doesn't seem like much, clearly not the light-years that would be needed to collide stars together. However, this happens inside stars too. What if that shock was enough to trigger supernovae? Actually, the mechanism is not the point of the question. Just suppose there is some mechanism like that.
The actual question is: is the shape of spiral galaxies consistent with the shape of gravitational waves that would be emitted by black hole mergers at the galaxy centres? Taking into account that the gravitational waves travel in the radial direction, while the galaxy is rotating? From a naive qualitative picture, it seems that the typical two spiral arms, the symmetry, the bars, the galaxy rotation curve could all be consistent with this idea. Or not?