Adding DRAM for non-gradient based sampling

Hello, I recently found pyMC and I am very impressed with it. However, much of the work I do relies on physics models that are run on codes that do not produce gradients. For these types of problems DRAM (Delayed Rejection Adaptive Metropolis), is a rather robust sampler that has been shown to be effective for the types of problems I work on. Is there any chance that DRAM might be added to the sampling offerings of pyMC?

Thanks for taking the time to read my request.