First the DNA is amplified. There are dozens of different algorithms for then analyzing the nucleotide sequences. Every living organism is a bit different from all other organisms. Obviously, a lot of DNA is shared among life forms (we are 60% the same as bananas; 99% the same as chimpanzees). So, the human genome project looks at the *other* genes to ascertain first that we're looking at a human. Each allele in the human genome is surrounded by nucleotides that are not part of the code - they tell us a lot, though (as they are not selected for and ride along in a population, helping us to distinguish ancestry).
So we look at the actual alleles. One would be the allele for hemoglobin which has about 500 variations worldwide. Most people have two different alleles there (one from mom, one from dad). All the similar alleles are looked at. It doesn't take long before we have an individual DNA analysis for a person. If we found 3 different alleles for this trait, it would mean that more than one individual was involved (having an extra chromosome is a serious problem and rarely results in normal physiology). Naturally, we'd want to see more than one example of this kind. Different algorithms are applied in different labs, but as the data comes in, we know how to string all the alleles together, because the human genome is well studied (we know the length and the basic codons that are essential to each allele - we look, again, at the uniquely human alleles, which are about 2700 alleles, each made up of thousands or hundreds of thousands of individual base pairs of nucleotides - you may have learned those as A,T, C and G).
One gene can have up to 200 million bases. Others have only a few hundred. We look hard at those long ones.
We now know that all organisms have mutations - and
at a higher rate than previously thought.
So, to make it brief: we look for alleles that are unique and we look for the complete genome through our knowledge of what an allele is and what so-called "junk DNA" is (the part that separates the alleles - although we now know that this DNA may actually provide certain instructions to the body). If we compare relatives, the average person is about 50% identical to their parents in overall allele function, but may have a single point mutation somewhere in these millions of bits of data. This person will pass that on to their own offspring - but a point mutation is likely to occur somewhere else. Many point mutations are neutral in the context of a long allele.
Siblings usually share 50% of those alleles. Half-siblings, about 25%. Two different humans will not have the exact same alleles. So looking for a second source of DNA means first making sure that the DNA Is completely mapped and that there are no more than 2 alleles per location. Three alleles that fit into a location (and remember, hundreds of thousands of alleles are known and in the algorithm already) is a sign that there's another contributor. Zero extra alleles means it all comes from one person - but even, then, there are second and third ways of checking to see that it's one individual.
Hope that helps. In Kohberger's case, I'd be so interested to see if he has any mutations in the area that controls the visual cortex (where VSS is thought to operate). Of the alleles that are unique to humans, a lot of them do have to do with the brain. Interestingly, alleles are used and reused in various combinations in many different body functions (so an unusual visual cortex allele might be associated with several different outcomes, in terms of physiological traits).
Certain regions of human DNA have such high variability (like the hemoglobin gene) that it's fairly easy to see whether just one human is involved; if it were only that one locus/allele, of course, the chances would be about 1 in 500 of uniqueness - but there are several other locations with even more alleles - even for eye color, there's not just one gene for brown eyes or blue eyes - there are way more for brown than for blue, but there are many for each of those; we also know exactly where to look within the context of bits of DNA for those alleles, even if the DNA is not in chromosomal form). So if we look at 2000 locations, with up to a power of 300-1000 (available alleles) we get a very large number: 2000 to the 300th power is huge). So, if even ONE stray allele (extra) is found in this area, one allele with a tiny difference from the two others expected - it's two people.
And the hemoglobin gene does not affect survival - each of those alleles is fine; but some are ancient (1 million years old - coming to us from a pre-human population) and some are relatively new ones (that is another way of estimating ancestry). Y chromosome is another interesting system, as it is a small, compact chromosome that is often found entire in a sample. Finding two different Y chromosomes in a sample means two different people.
HTH. I know it's long. What the computer does is insert its "knowledge" of both "junk" DNA and the human genome into the analysis of what is found on the swab. Most times, long bits of a chromosome are still intact, but even if not, the computer recognizes which allele goes where, as they serve no purpose in a living organism other than to keep the organism alive and to build the body structure we see with our eyes. Each location is slightly different, IOW; and then the alleles are different too.