Ultimately, it is your responsibility to check the information that you use in your paper. If the data that went into your paper was wrong (even if you or your co-authors didn't take that data), then you have some obligation to correct that error. It's may not be your fault, since obviously there are finite resources to check things and you may not even have had all the details necessary to reproduce the data you used, but it is now your responsibility as a scientist to set the record straight about the consequences for your work. (It is also counterproductive to worry too much about whose fault an unintentional error is.)
Deciding what action to take about your paper is similar to deciding what to do about any other kind of error. For a minor error that does not affect the main conclusions, but changes some details, you might add a comment to a future paper updating and correcting the result. For a major error that changes a key result significantly, you might submit some kind of erratum. For a truly egregious problem that invalidates the premise of what you did or your main conclusions, you might need to retract the paper.
In addition to correcting your paper, you should also try to fix the underlying problem with the source of the data. It is a good idea to contact the authors of the original paper so they are aware. If this is a minor issue, you might want to add a comment in one of your upcoming papers where you point out the issue with the data in the Supplemental Material. If this is a major problem, you could imagine publishing a paper refuting the data in the original paper (although I doubt this would be reasonable to do in this case, based on your description).
These things happen. There are certainly many cases of an error in one paper being carried on and used in future papers. The main thing is to try to clean things up, so there is a record correcting the problem in the literature.
