Currently, dpdusage prints unused lemmas in an unspecified order that can change unpredictably when including another module in the analysis. In my current use-case (moving a theorem and everything only used to prove this theorem to a separate file) I was interested in the diff between the unused lemmas with and without the extra file. This required a sorted output without the Info line.
A typical invocation of dpdusage might thus look as follows:
dpdusage graph.dpd | grep -v '^Info:' | sed 's/\([^ :]*\):\([^\t]*\).*$/\1.\2/' | sort
(the use of . over : simplyfies copying lemma names into a "whitelist definition" as a hack until #14 is merged)