1. Estimating the Rate of Gene Conversion on Human Chromosome 21
Badri Padhukasahasram, Paul Marjoram, Magnus Nordborg 
The American Journal of Human Genetics 
Volume 75, Issue 3, Pages (September 2004)
DOI: /423451
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

2. Figure 1
A likelihood surface of chromosome 21 data that is based on the fraction of short-range (0–5 kb) incompatible SNP pairs alone. ρ denotes the crossing-over rate, γ denotes the gene-conversion rate, and L(ρ,γ) is the product of the likelihood and an arbitrary constant. Note that estimates of gene conversion decrease rapidly with increasing crossing-over rates and eventually reach zero. A single pairwise summary cannot distinguish between the effects of gene conversion and those of crossing over.
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

3. Figure 2
Expected values of p(a) and p(b), from simulations under models with uniform crossing over alone (γ=0) and uniform gene conversion alone (ρ=0) at differing rates. Expectations were calculated from 1,000 simulations of 200-kb sequences, with θ=140. For models with gene conversion alone, the mean conversion-tract length was fixed at L=500 bp.
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

4. Figure 3
The distribution of MLEs of gene conversion ( γˆ) for 1,000 data sets, each simulated at uniform crossing-over rate ρ=40 and uniform gene conversion rate γ=40 (A) and at nonuniform crossing over ρ=40 and uniform gene conversion rate γ=40 (B).
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

5. Figure 4
The likelihood surface of chromosome 21 data based on p(a) and p(b), for a mean tract length L=500 bp. Likelihoods were calculated from 8,000 simulations of 200-kb sequences for different rates of crossing over (ρ=0–160) and gene conversion (γ=0–300). L(ρ,γ) roduct of the likelihood and an arbitrary constant. The peak is seen at ρ=80 and γ=125.
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

6. Figure 5
The likelihood surface of chromosome 21 data based on p(a) and p(b), for a mean tract length L=50 bp. Likelihoods were calculated from 2,500 simulations of 200-kb sequences for different rates of crossing over (ρ=0–160) and gene conversion (γ=0–1,500). L(ρ,γ) denotes the product of the likelihood and an arbitrary constant (1,500). The peak is seen at ρ=80 and γ=750.
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions

7. Figure 6
The MLEs of gene conversion ( γˆ) and crossing over ( ρˆ) for overlapping 2-Mb windows along chromosome 21. The X-axis denotes the location corresponding to the center of the 2-Mb windows, and the error bars indicate ∼95% credible intervals. The horizontal line denotes the chromosomal average of the parameters.
The American Journal of Human Genetics  , DOI: ( /423451)
Copyright © 2004 The American Society of Human Genetics Terms and Conditions