You shouldn't expect DNA sequences to line up across sp. or Kingdoms. Predicted amino-acid sequences will be a better indicator for sequence similarity, but keep in mind that even protein sequences with 10% identity can have the same fold. ANTISmash, SMURF, and NapDos use Stachelhaus code to predict the substrate specificity of Adenylation domains in the NRPS. The code is derived from many different experimental analysis of adenylation activation domains, but the fungal NRPSs are underrepresented. For instance, if Adenylation domain 1 from a predicted NRPS operon in ANTISmash suggests a Leucine then it might be any volumetrically and non-polar similar residue. Since fungal NRPS often incorporate Isovaline and 2-amino-isobutyric acid (think about the peptaibols) you should consider that these residues might be present in the final product of the predicted operon. The rules are more clearly defined for the NRPSs due to the biosynthetic modularity and wealth of literature.

The PKS and terpenes are more difficult to examine in silico. Their rules are less clearly defined due to the inherent complexity of their biosynthetic processes. However, if you have fractionated the extracts of the fungi in question then you can look through the bioinformatic analysis for a biosynthetic scheme that makes since in the context of an isolated product that you purified from the fungus in question. Anytime I examine biosynthetic operons from sequence I compare with experimentally identified and confirmed natural product databases (SciFinder, Dictionary of Natural Products, MarineLit) using the source organism genus ors species as a search vector.