Chimpanzee Contigs and the Human Genome: 84% Similar?

[T_aquaticus]

@John_Harshman

Aside from the problem already mentioned, there’s the problem that a single indel of 17 bases counts as (at least) 17 differences, when it should count as one difference at most. The common figure of 98.7% sequence identity is for aligned sequences, which doesn’t count indels.

As long as the differences are accurately described I don’t have a problem with using a straight base to base comparison. The chimp genome paper has the figures we need for comparisons to Tomkins’ work.

The analysis of modest-sized insertions reveals ∼32 Mb of human-specific sequence and ∼35 Mb of chimpanzee-specific sequence, contained in ∼5 million events in each species (Supplementary Information ‘Genome evolution’ and Supplementary Fig. S5).
Chimp genome paper

67 million bases of unique DNA in a 3 billion base genome is an additional ~2% difference stacked on top of the 1-2% difference due to substitutions. This puts the total at around 96-97% for a base to base comparison.

@John_Harshman

Now, I’m not sure Tomkins even gets that far. He’s done a lot of weird things before. I believe that one of his distance measures has just counted the percentage of randomly chosen 30-base segments identical between species — no indels, no base changes — as a distance measure, which is odd in itself and really can’t be compared to any other sort of measure.

As mentioned by others, he used an ungapped analysis which means that a 300 bp comparison that differs by a single indel in the middle of the sequence would be counted as ~75% similar even though it differs by a single base.

At least this time around Tomkins appears to be including gaps, but time will tell if they were correctly accounted for.