In 2013, geneticist Eugene McCarthy published a theory (link) about human origins, which I'll summarize:
A female bonobo mated with a male pig, and their offspring interbred with bonobos for several generations. The resulting mostly-bonobo, part-pig hybrid is the ancestor of Homo, explaining key anatomical differences between humans and other primates.
McCarthy was generally mocked for the theory, (e.g. see Richard Dawkin's tweet from 2013).
A couple months ago, McCarthy posted a followup study analyzing DNA instead of just anatomical traits. A quote from the conclusion:
So, the results of this study indicate that essentially all the nucleotide positions that differ in humans and bonobos, are the same in humans and pigs.
Will this finding be checked by anyone else, and will they find the same result?
Resolves YES if someone else compares the triad of bonobo (or chimp)/pig/human DNA and they get essentially the same result. Resolves NO if no one else does the study or if they get pretty different results. Since who counts as a replicator, and how different the results need to be before they are really different is a judgement call, I will not vote on this poll. I hope there will be consensus on how this should be resolved when the time comes.
I found a human, pig, and bonobo genome online and built a little flask app that compares them. I sampled tiny chunks of human DNA at random (40 characters long) and searched for matches in the pig and bonobo DNA. I allowed for 5 differences, just like the paper.
I categorized the results into 4 buckets:
Just human
Human-bonobo
Human-pig-bonobo
Human-pig
I was only able to run 900 tests last night (it takes a while to search the whole genome and I was only only 3 processes due to memory issues that I need to fix). Here were my results:
Just human (287, 31.8%)
Human-bonobo (550, 61.1%)
Human-pig-bonobo (55, 6.1%)
Human-pig (8, 0.8%)
I am not sure why I got 31% just human. That doesn't match with 98% of DNA being shared between humans and bonobos. But I didn't do the pre-filtering that McCarthy did in the paper. I just dumped the raw genomes into the process and I left the Ns in there (I did remove the distinction between upper/lower case).
If we count the 31% of "just human" as bad somehow (which a lot of it has to be since humans do share 98% of their DNA with bonobos), then the percentage of human-pig goes up from 0.8% to 1.12%, which is pretty close to the 1.3% McCarthy got.
I want to optimize my code to be able to run more samples, think about why I'm getting such a high hit rate for just human, and so on. But I got excited about how it sort of matches his results this morning and wanted to share what I got.
I uploaded the code here: https://github.com/Jeremiah-England/hybrid-hypothesis-replication
It's pretty messy now and the README isn't designed for onboarding. I'll comment back here once I have it more user-friendly in case anyone want to try it for themselves.
@JeremiahEngland Note, I'm not planning to count whatever I end up doing as a replication at this point. And I would wait for ~consensus here before doing so anyway.
Last night I ran the comparison again. This time on more cores and with an extra genome in the mix: a cow.
The comparison logic was the same, take 40 character long strings out of a human genome and try to find matches within 5 characters in the other genomes. With three other genomes to compare to, there were 9 categories instead of 4.
Here were my results:
total: 1512
Human-bonobo: 813
Human only: 550
Human-cow-bonobo-pig: 71
Human-bonobo-pig: 30
Human-bonobo-cow: 22
Human-cow: 11
Human-pig: 8
Human-cow-pig: 7
As you can see, the ratio of what the human-pig combination shared that the bonobo genome did not was similar to my previous result. However, the cow came through with about the same number of extra matches to human DNA outside of bonobo as the pig.
I think tonight I'll try doing the direction of comparison the same way that McCarthy did, with pig as the source and human/bonobo being matched against.
I also need to figure out why my match rate between pig/human is not closer to the 50% that he got in the paper.
https://x.com/Macroevo/status/1833233884613619871
"I'll be submitting this preprint for peer review in a few days. If anyone sees a problem that needs to be fixed, please let me know. Thanks."
@Slackhammer After reading his website and the paper, I personally think this is more likely than not. But I am less confident that someone else will try to replicate it.
Other links:
Tim Tyler's blog post: https://on-memetics.blogspot.com/2024/08/part-primate-part-pig.html
Eugene McCarthy's twitter: https://twitter.com/Macroevo