Below is my poster and 10 minute audio overview for the 2022 Population, Evolutionary, and Quantitative Genetics conference. Transcripts and figure descriptions are also provided for accessibility.
If you have any questions, feel free to reach out over email [cad17 (at) g.ucla.edu] or Twitter @tinadelcarpio.
Poster Presentation Audio
Figure Descriptions
Transcript of Poster Presentation Audio
(00:01):
Hi, my name is Christina Del Carpio, and I’m a PhD candidate in the Lohmueller lab at the university of California, Los Angeles. And my research I’m presenting on is a dissertation chapter about the evolutionary patterns of recombination rates in gray wolves and dogs. So meiotic recombination is a process that we know varies by species populations, individuals, and sexes. Uh, and we also know that there’s a gene called PRDM9, that controls recombination rates and location in most mammals, but its function was lost early on in the can lineage. There has been previous work comparing recombination rates and locations between dogs and wolves, but it was limited to less than 20 gene regions. So for our work, uh, we used high coverage, whole genome data from a population of wolves in the Canadian Arctic, and also, uh, from pug, uh, sequences using that data.
(01:02):
I inferred the demography with SMC plus plus, and then using a program called pyrho. I estimated the recombination rate or R additionally, I also estimated the recombination rate per chromosome of this empirical data, the sequencing data using an SMC demography with a constant population size with the most recent population size as a constant any, uh, additionally I worked with my undergrad, uh, research assistant to simulate canid genomes using program msprime. Um, and those simulations were modeled under the demography. I inferred from the empirical data. Then with the simulations, we followed the same pipeline of using SMC plus plus to estimate demography and then pyrho to infer recombination, but also tried different, um, scenarios for the simulations, including inferring r again with this constant population size, um, from the most recent population size that SMC estimated. Um, and then we also did a test where we added runs of homozygosity or sort of these long regions without any genetic variation to the simulated dog genomes. And then in turn inferred r, and our results, we found that wolves have a recombination rate that is about 18 times higher than domestic dogs. So if you look at figure two in the middle here, we have the recombination rate for chromosome. One for wolves and pugs represented. The Y axis is the recombination rate per base pair per generation. And then the X axis is the chromosome position along chromosome one. And in this orange color, we have wolves represented and in blue we have pugs represented
(02:58):
And you can see there’s some, uh, alignment sometimes where these recombination values peak, but overall it’s very striking that the recombination values for wolves are noticeably higher than it is for pugs. And so then we took to these simulations to test, is this a true biological phenomenon or possibly just an artifact of our inference methods, but based on the simulations we’ve done so far, this difference is not explained by differences in demography or in heterozygosity. So looking at figure three, uh, we have the recombination rates for simulated Wolf and pug genomes, and again, on the Y axis recombination rate. And then on the X axis, we have the different scenarios, um, that my undergraduate research assistant and I tested. Um, so first looking at the, at the first two, um, violin plots we can see on the left is the Wolf sample on the right is the dog sample.
(04:03):
And this is under the infer like using the regular inferred demography from SMC plus plus, um, you can see there’s a dotted line at the top of this figure. That is the true recombination rate that was used in these simulations of one times 10 to the negative eight. And you can see that for both pugs and wolves that we actually underestimate the true recombination rate, but comparing against dogs and wolves. Uh, it seems like the median values and the range of values are incredibly similar and just simulating this genetic data under different demographies. Doesn’t reproduce this 18 fold difference. And now if we jump to the next two violin plots, these are where we inferred are using a constant, any of the most recent population size. And here the, the estimated r is much closer to the true value for both pugs and wolves. But again, we see that the values for pugs and wolves are very similar and have significant overlap in their range.
(05:07):
Um, so again, we don’t see this explaining the 18 fold difference. We see in the empirical data of recombination rates. And then finally our last violin plot here is from the dog genomes where RoHS were added. So the RHS were added based on what we know patterns of RoHS or this lack of genetic diversity looks like in domestic dogs. And if you compare that against the inferred demography of the dogs or that second violin plot within this bigger plot, you can see that those values are very similar, um, to one another, the medians are similar and the range overlaps. So adding, uh, pattern of RoHS that matches the domestic dog genome also doesn’t significantly shift the inference of the dog genome
(05:56):
Enough, to explain this difference between dogs and wolves, and then moving on to some additional results, uh, here in figure four, you can see the demography itself that I inferred with SMC plus plus. So again, uh, wolves are in this orange color and pugs are in this blue color and we have on the X axis, the effective population size, uh, or excuse me, on the Y axis effective population size. And the X axis is generations ago. So from right to left, it goes from past to present. Um, and you can see that some of these more historical sizes, um, come very close to one another for pugs and wolves, but we also see a dramatic drop, for example, in pugs that sort of aligns with the domestication of dogs. Um, and we can see that there are differences in the demography between dogs and wolves, even though, um, wolves are a progenitor of dogs, but because of this bottleneck from, um, the actual breeding of dogs and the separation of these two, there are some differences.
(07:02):
Um, and then just focusing in a little bit more on how changing the demography we use to infer our, um, does not impact the pattern of Wolf recombination being significantly higher than dog recombination. Uh, we can look here at figure five. Um, so again, recombination rate is on the Y axis, and then we have species on the X axis and the panel on the left represents using the inferred SMC plus plus demography and the pyrho analysis. And then the panel on the right represents using the constant population size of the most recent population size. Um, and so first, if we focus on this panel on the left, uh, each of the points in this box plot represents the mean our value of a given chromosome, um, within the dog or Wolf genome. And you can see, again, this pattern just represented a different way, but similar pattern from figure two where the mean recombination rates for wolves are about 18 times higher than dogs.
(08:06):
And when we switch to using this constant any, uh, of the most recent population size on the right panel, we see a very similar trend. The biggest difference is that the Wolf data has significantly less variance, which makes sense. If we look back at figure four with the demographic models, you can see that wolves have a slightly more complex or basically, uh, more variable, um, demography than pugs. And so by simplifying that Wolf demography to a single constant value, it’s not surprising that in turn, the analysis has less has a result that has less variance. So overall we found, again, that wolves appear to have a higher recombination rate than dogs and our simulations so far show
(08:54):
That that may be a true biological phenomenon, but in our next steps, we’re also going to, uh, look at how changes in mutation rate impacts our inferences of r um, especially because the mutation rate we used was inferred specifically in a population of wolves. Um, we’ve also started some analysis on more dog breeds. So we’ll also be including those analyses. Um, and then my next dissertation chapter will be focused on a pedigree method of inferring recombination using the population of almost 400 wolves in Yellowstone national park. Um, thank you to all of my co-authors and all the members of the Lohmueller lab for their assistance. Thanks to my cats, Tuca, Jem and my partner tests for their support. Um, funding has been provided by an NIH NHGRI T 32 training grant as well as the NSF, G R F P fellowship. Um, and thank you so much for taking the time to see my poster. If you have any questions, please feel free to reach out, uh, via email, um, or Twitter. My contact info is listed at the bottom of my poster. Thank you so much.
Poster Text for Screen Readers
The Evolutionary Patterns of Recombination Rates in North American Gray Wolves (Canis lupus) and Domestic Dogs (C. familiaris)
Christina A. Del Carpio1, Maria Izabel Cavassim Alves1, Pedro Angel Perez1, Robert K. Wayne1, Kirk E. Lohmuller1
1University of California, Los Angeles
Background
•Recombination varies by species, populations, individuals, and sexes1
•PRDM9 controls recombination in most mammals, but function was lost early in the canid lineage2
•Comparison of dogs and wolves limited to <20 gene regions3
Methods
•Data: high coverage WGS data from population of wolves4 (n=15) and pugs5 (n=14)
•Inferred demography with SMC++ (Fig 4)
•Estimated recombination (r) using pyrho (Fig 2)
•Estimated r per chromosome of empirical data using SMC++ demography and constant Ne of most recent population size (Fig 5)
•Simulated canid genomes with msprime with inferred demographies (Fig 1)
•Followed pipeline for empirical data (Fig 3, colms 1&2)
•Inferred r using a constant Ne of most recent population size from empirical data (Fig 3, coms 3&4)
•Added runs of homozygosity (ROHs) to simulated dog genome and inferred r (Fig 3, colm 5)
[image: figure 1]
Wolves have a recombination rate ~18x higher than domestic dogs
[image: figure 2]
This is not explained by differences in demography or heterozygosity.
[image: figure 3]
Additional Results
Inferred Demography Does Vary Between Wolves and Dogs
[image: figure 4]
Changing Demography Used to Infer r Does Not Affect Pattern of Wolf Recombination >> Dog Recombination
[image: figure 5]
Next Steps
•Simulating how changes in μ impact inference of r
•Inferring r in more dog breeds
•Pedigree based inference of r in Yellowstone wolves
Acknowledgements
Thank you to all my co-authors and members of the Lohmueller Lab. Thanks for support from Tuca, Jem, and Tess. Funding provided by NIH NHGRI T-32 training grant and NSF GRFP.
References
[1] Kong, A., et al. (2010). Fine-scale recombination rate differences between sexes, populations and individuals. Nature. [2] Axelsson, E., et al. (2012). Death of PRDM9 coincides with stabilization of the recombination landscape in the dog genome. Genome research. [3] Muñoz-Fuentes, V., et al. (2015) Molecular Biology and Evolution. [4]Robinson, J. A., et al. (2022). The critically endangered vaquita is not doomed to extinction by inbreeding depression. Science. [5] Marchant, T. W., et al. (2017). Canine Brachycephaly Is Associated with a Retrotransposon-Mediated Missplicing of SMOC2. Current Biology. [6] Koch, E. M., et al. (2019). De Novo Mutation Rate Estimation in Wolves of Known Pedigree. Molecular Biology and Evolution.
Contact & Further Details
Tina Del Carpio | Email: cad17@g.ucla.edu | Twitter: @tinadelcarpio | Audio: tinadelcarpio.com/PEQG
Tina Del Carpio 