Molecular biologists could be reporting false experimental data because they are being overly influenced by previous findings.
Molecular biologists could inadvertently be reporting false experimental results because they are being overly influenced by previous findings, report a team of US bioinformaticians.
The team, led by Andrey Rzhetsky from Columbia University, US, came to the conclusion after statistically analysing 3.3 million experimental findings on molecular interactions, reported in 78 different journals.
Each finding consisted of a positive or negative statement about the interaction between two molecules, such as ’protein A activates gene B’. Rzhetsky and colleagues coded positive statements as 1 and negative statements as 0, and then recorded in chronological order the positive and negative statements made by different researchers about the same two molecules.
To interpret these chains of 0s and 1s, the researchers developed a probabilistic model that generated five distinct patterns of data based on the level of influence between successive experiments. These extended from a ’trust nobody’ pattern, where scientists ignored other experimental results, to a ’super-conformism’ pattern, where scientists were strongly influenced by previous research.
Rzhetsky and his team found that all five patterns could be detected in their real-world data, but that the most prevalent was ’mild scepticism’. This corresponds to the situation where scientists read other research papers but tend to trust their own experimental results more than that of their contemporaries.
Although this is a fairly ’common-sense’ finding, Rzhetsky also found that the influence from previously published work was still strong enough to prevent researchers from reaching a ’correct’ result. He suggests that this problem could be alleviated by ’restructuring the publication process or introducing a means of independent benchmarking of published results’.
The findings are fascinating, said Simon Colton, a lecturer in bioinformatics at Imperial College, London. ’While it isn’t quite at the level of the football terraces, there is group behaviour evident in scientific research,’ said Colton. Nevertheless, he predicts that scrutiny from the pharmaceutical industry and society at large will prevent molecular biologists from going too far down the wrong path. ’While I tend to accept [the researchers’] findings and interpretations, I’m not too concerned about the state of biomedical research,’ he told Chemistry World.
et alProc. Natl. Acad. Sci. USA (DOI: 10.1073/pnas.0600591103)