Given the very high number of variables being probed, it's not surprising some differences are found. No replication samples are tested, so my prior expectation would be that these are spurious, until proven otherwise

 — @WiringTheBrain