Firstly, I would be more inclined to take this finding seriously if the genetic material had been obtained from autistic children who had been exposed to psychoactive medications in utero rather than from a sample with "idiopathic" autism. Surely "idiopathic" means there is no reason to consider these people similar currently. Why not compare the fish genes with genes from children whose parents were known to be taking psychoactive substances during conception & gestation?
Also, although we are phyllogenetically, distantly in line with fish (especially as embryos), our '"behaviours"' are grossly different, so I tend to think similar genetic material could not lead to similar behaviour.
Last, I understand that substances act at different embryological stages in human brain development (so some drugs are excluded during the first/third trimesters) so that the number of "layers" of enzymes & biochemical reactions that foreign substances must "pass" before they become embedded in genetic material must be a lot greater in humans than in fish.
In my opinion the link between gene anomalies & autistic phenotype is not well-established in most cases except family clusters. Therefore, I believe the extrapolation from the present fish findings to autistic humans is basically unwarranted in the case of psychoactive substances.