There are a number of questions about accessing the data corresponding to publications. In the recent years, many journals have improved their data availability policies and this is less of an issue. On the other hand, I do not find much about the extent of exploiting these data that is thought acceptable.
Firstly, I believe most scientific communities accept that other's data are replotted along with similar data acquired by the authors (showing repeatability) or along with similar data from yet other groups (meta-analysis type of figure). Still, I have the impression that in many communities people rather plot only their own data and comment that this is the same behavior as reported by so & so. So is there a policy? Should one ask for author's permission? Should one specify exactly what data of theirs will be shown along with what other data?
Then, there's the question of data re-analysis. For instance, say authors have published the time evolution of several variables of interest, and Methods make clear that these were acquired simultaneously on the same sample. Is it OK e.g. to publish separately from them a figure which plots one variable from their dataset against another? I guess the answer to this question can be yes or no, depending on the way this is done. I expect that in most cases, having this plot with only their data as my Fig 1, possibly overlaid with a model prediction, and being the main result of my paper, will be judged problematic. Is it? If yes, what is the red line between acceptable re-analysis (e.g. I've already seen data re-plotted in log scale, but the figures with this data were in supplementary materials only). And of course, same questions as above apply.