Thanks so much!It looks promising.
If I have two or more derivative data and want to combine all these information. Can I just do MRSAD individually for each dataset, and combine them?

Short answer: much of this is already automated in AutoSol, and the model phases can be used for methods other than SAD. You'd still run AutoMR to get the model phases, then run AutoSol with all derivatives at once plus model phases. I'm not sure whether it's better to do it as an MIR(AS) experiment, with your native data and the derivatives, or multiple SAD experiments (without native data). Both should be (mostly) straightforward to set up.

However, the quality of the final phases still depends on the quality of the derivative data. If you already have decent phases it is often very easy to find heavy atoms by eye in the anomalous (or isomorphous, I guess) difference maps, but this doesn't always mean that you can calculate good phases using those sites. It might be a good idea to try running Xtriage on each of the derivative datasets first to assess the anomalous signal.

-Nat