The lizard DNA won't bind to the beads in the Illumina bead chip microarray. (I used to help design them; 23andme buys them from ILMN) The swab was contaminated with human DNA. (sorry for the spoiler)
Arrays are cheap and old technology. It's like taking a sample of a word from every page in a decent sized phone book. But super duper cheap. If you really want to do genetic analysis these days you'd use NGS (Next Generation Sequencing) capable of reading every word in the phone book. The step after that would be variant calling (typically) where your phone book is compared to a reference phone book and differences are noted. (outputs a VCF file if you care to dig into that)
The reason the lizard DNA isn't legit is the chip is covered in short sequences of human dna. Only complementary DNA (other strand of human DNA) would bind to it. Now if it were the case that 'some' of the lizard DNA just happened to bind it wouldn't be much and would fail QC. You'd at least need most of the beads hybridized with DNA to pass QC, thus, it's human DNA. Or the whole thing might just be BS.
There are dropouts in microarrays so that would explain your situation. A certain amount is tolerated. Same with NGS sequencing but it's called missing coverage - you just don't get a read over certain areas.. However good sequencing panel designs are aware of difficult regions and enrich for them.
Someday these ancestry companies will switch to sequencing but it's still ~$500-1K per genome. Most people wouldn't pay for that.
#humor