here are two basic approaches to genome maps: the physical and the genetic. The sequence is the ultimate physical map, based on the actual content of the genome. At #6 is a new and more detailed kind of genetic map, which measures the probability of recombination along each chromosome. Lead author Augustine Kong and colleagues at deCODE genetics (including deCODE’s director and co-founder, Kari Stefansson) in Reykjavik, Iceland, have published a map that is at least five times more accurate than the previous best map, the so-called Marshfield Map, according to that map’s draftsman James Weber. deCODE genetics was established as a private company to profit from Iceland’s well-documented family histories, but it has made its recombination map publicly available.

Chromosomes, of course, come in pairs, and one in each pair is inherited from each parent. But the chromosomes are not inherited entire, like socks. Rather, portions are swapped between the two parental chromosomes during the meiotic cell division that precedes formation of eggs and sperm. It is this swapping, or recombination, that shuffles the genes and throws up new combinations for evolution to favor or discard.

The deCODE scientists recruited 146 Icelandic families and looked at the placement of more than 5,000 genetic markers in the sibs and one or both parents. They were looking specifically for recombination events, where markers that were on the same chromosome in one parent had become separated in the offspring. Because the resulting recombination map is derived completely independently of the physical sequence, the two offer useful checks on each other. There were several discrepancies between the order of markers as revealed by the recombination map and the order of those markers on the sequence, but, as much of the sequence was still in draft form when Kong’s paper was published, that is acceptable. Three finished chromosomes showed no discrepancies, but some may turn up as other chromosomes are completed.

One of the most important consequences of the new map is that it makes studies of genetic linkage, in which the presence of a disease is correlated with the presence of genetic markers, more accurate. The unit of recombination is the centiMorgan (cM); 1 cM is roughly equal to 1% recombination. With the Marshfield Map, the 95% confidence interval was 0.1-3.6 cM. In other words, for an observed 1% recombination there is a 95% chance that the true recombination rate lies between 0.1% and 3.6%. The new map brings the interval down to 0.5-1.8 cM, which will make it easier to locate genetic markers associated with disease, especially for weaker associations.

The high-resolution recombination map has provided insights into the fine structure of recombination. For example, shorter chromosomes have higher recombination rates than longer ones, the average rate for 21 and 22 (the shortest) being twice the rate for 1 and 2 (the longest). Rates also vary along the chromosomes, with recombination more likely at the ends than at the centromere. Along each chromosome there are distinct regions of high and low recombination. The Marshfield Map had located 19 so-called "deserts," where the rate of recombination is very low, and 12 "jungles" where the rate is very high. Kong and colleagues identified only 8 of the 19 deserts, but all of the jungles, and they hypothesize that the discrepancy reflects the relatively small dataset of the Marshfield Map.

Recombination depends on the chromosomes of a pair coming together and aligning, which in turn suggests that the detailed sequence where crossing over occurs will be important. Kong and colleagues plotted correlations between recombination rates and sequence, and found much higher correlations than previous studies. Three sequence parameters—CpG motifs, GC content and poly(A)/poly(T)—explained about 32% of the variance in recombination rate along the chromosomes. Along with the higher accuracy of the map, the ability to predict recombination rates based on sequence should make genetic linkage studies more informative.

The Icelandic data contain several puzzles for evolutionary biologists. One is that the recombination rate for women is roughly 1.65 times higher than for men. There are also regions along the chromosomes where recombination is high for men and lower for women, and vice versa. And some families have higher rates of maternal recombination than others. If recombination provides the raw material for evolution, could this mean that women contribute more than men to evolution, and some women more than others? The differences between men and women, and among individual women, suggest that factors other than the sequence come into play in determining the rate of recombination. In maize there seems to be a gene that can modify the rate of recombination. Might there be such a gene in humans too?