@glipsnort, @evograd, and @davecarlson, do you have any datasets that show the recombination breakpoints across the genome for individuals, probably using their parent SNP maps to figure out where recombination took place?
Not for humans, no. Malaria parasites, yes.
So what do you think the distribution of recombination breakpoints is? How would you mathatically model it. My attempt at starting rules:
- At least one crossover per chromosome.
- Perhaps up to 3 crossovers per chromosome, also more likely on large chromosomes.
- Self avoidance of crossovers, so no two are very close together.
- Totally number across a is about 33.
Do you agree with this? Would you change anything?
This paper might be of interest:
From your data is this happening only at the gene level or are exons sometimes getting mixed?
This one is based on 25k more meioses: https://science.sciencemag.org/content/363/6425/eaau1043. Newer! Shinier!
Depends on what you’re modeling. If you’re interested in fairly short time scales and large chunks of chromosomes, it sounds reasonable. I’d put one on each chromosome, pick a Poisson-random number of additional events (whatever gets you up to the sex-averaged recombination rate (3400 centimorgans for the autosomes) and distribute them with probability based on chromosome length (physical length if you’re lazy, genetic length if you’re feeling punctilious). This assumes you don’t care about male/female differences.
Only if you’re dealing with thousands of generations does the nonuniformity of recombination really start to matter, in which case using an actual genetic map might become necessary.
Okay. How much avoidance? In a single generation, we shouldn’t have adjacent recombination, right?
Right, there is interference. I don’t know the details of estimates, but the range seems to be some tens of centimorgans (or Mb). Say 50 Mb?
About what percentage of a chromosome is that? Roughly.
50 Mb should work out to (roughly) anywhere between 1/2 and 1/4 of the human chromosomes.
Chromosome 1 is 246 Mb and chromosome 22 is 49 Mb - so a little different for the largest and smallest of the chromosomes.
Interference isn’t complete, though. You could reduce the probability within a radius – you’d have to look into the research on how completely recombination is suppressed. It could start to get complicated.
Whether interference is important at all again depends on why you’re doing the modeling. For some applications, gene conversion would be a more important factor to include.
This is for an educational visualization. It does not need to precise to multiple digits, but accurate enough that key visual patterns observable in the display are not misleading.