I didn't comment on this study when it came out in 2009, but as I'm reviewing some materials I thought it worth taking down a note. Carles Lalueza-Fox and colleagues [1] intensively sequenced the TAS2R38 gene in a Neandertal bone flake from El Sidrón cave. This is a gene in humans that enhances bitter taste perception, and is responsible for the classic taster/nontaster polymorphism for the substance PTC. Some 1.5 million years ago, a mutant version of this gene arose with less sensitivity to bitter substances; both the high-sensitivity and low-sensitivity versions exist in human populations today. People who carry two copies of the less active allele are often called "non-tasters", in contrast to "tasters" who have at least one copy of the more active allele.
Part of the difficulty of studying Neandertal genetic information is the low coverage of the genome reads available to date. Some parts of the sequence are not covered at all, and only a small fraction of sites in the genome are covered multiple times. If we want to study polymorphisms in a single Neandertal individual, we are limited to those areas with high read coverage, and even then we shouldn't put much confidence in them.
For TAS2R38, Lalueza-Fox and colleagues did much deeper sampling of a single relevant site by PCR amplification. They ended with thousands of reads of the site they targeted:
A total of 4307 sequences were generated for the TAS2R38 gene F142-R166 fragment (figure 1). Of the total, 2391 (55.51%) showed a C in nucleotide position 145, corresponding to a proline amino acid (taster haplotype), and 1916 showed a G (44.49%), corresponding to an alanine amino acid (non-taster haplotype). Three clones show singleton C to T or G to A substitutions that are the most common form of postmortem DNA damage (Briggs et al. 2007). The main researcher involved in the laboratory analysis (C.L.-F.) is proline homozygous. All the Y-chromosome sequences identified (n = 141) showed the ancestral allele and, thus, no male contamination of European origin could be detected in this amplification.
That's a pretty good argument in favor of this individual having the polymorphism in question. There is of course a high probability that all these reads actually come from a very small number of template molecules, so it's not as convincing as it might look. But it's a picture of the kind of work involved in confirming polymorphisms from ancient sequence data. We will probably be reasonably confident when we have read coverage up to 15-20x coverage for most loci.
Then we'll just have to worry about phasing. Maybe El Sidrón, with its related individuals, will turn out to be the perfect site for genomics.