Natural Selection Studies Based on Bad Statistics
Hundreds of studies claiming to show natural selection may be wrong, say scientists from Penn State and Japan. PhysOrg reported today that “several statistical methods commonly used by biologists to detect natural selection at the molecular level tend to produce incorrect results.” Many studies of human evolution have relied on these flawed methods. If the methods were wrong, the conclusions are unreliable. “Of course, we would never say that natural selection is not happening, but we are saying that these statistical methods can lead scientists to make erroneous inferences,” said Masatoshi Nei (Penn State), the leader of the analysis.
Associating natural selection with genetic changes that result in a different amino acid being substituted in a protein is a dubious assumption. “Actually, the majority of amino acid substitutions do not lead to functional changes, and the adaptive change of a protein often occurs by a rare amino acid substitution,” Nei said. “For this reason, statistical methods may give erroneous conclusions.”
The authors re-examined a 2007 paper by Yokoyama (see 09/05/2008) that Austin Hughes (U of South Carolina) had boasted last year represented the right way to infer positive selection. Hughes had himself ruled out the validity of statistics used to show natural selection, saying, “Thousands of papers are published every year claiming evidence of adaptive evolution on the basis of computational analyses alone, with no evidence whatsoever regarding the phenotypic effects of allegedly adaptive mutations.” He had pointed to the Yokoyama paper as an exception – a study “solidly grounded in biology.” Now, these scientists have apparently shown that even that case was flawed. There was no correlation between predicted sites of selection and those determined by experiment – Yokoyama had found false positives, they claim.
What would be required to test for natural selection more accurately? Nei’s team said that scientists should pair statistical data with experimental data whenever possible. This, however, is difficult and costly. The article ended, “Scientists usually do not use experimental data because such experiments can be difficult to conduct and because they are very time-consuming.”
The Penn State press release is visible at the Penn State Live news website. The paper appeared on PNAS Early Edition on April 1, 2009.1
1. Nozawa, Suzuki and Nei, “Reliabilities of identifying positive selection by the branch-site and the site-prediction methods,” Proceedings of the National Academy of Sciences, published online before print April 1, 2009, doi: 10.1073/pnas.0901855106.
Tell this to Texas high school students. Remember the pile of papers the Darwin Party stacked at the Dover trial showing evidence for evolution? It was all fluff by lazy scientists unwilling to pay the price to get valid scientific evidence for Charlie’s grand myth. Cut away the bad statistics, the storytelling and the assumption of evolution offered as evidence for evolution, and the stack would disappear. Notice that to protect their safety, these Penn State folks had to declare that “Of course, we would never say” that natural selection is an overhyped, vapid process. It’s time to call the Darwinist’s bluff. Show us the studies that really establish Charlie’s Stuff Happens Law can create wings out of slime. Creationists have a bigger pile of evidence for their view. The universe.