Title: The Emperor’s New Clothes: The Many Sins and Crimes of Biomedical Big-Data “Science”
Date: 17/12/2019 
Time: 12:00 
Location : Sfadia Auditorium, Multi-Purpose Building

We are very pleased to invite you to the Natural Sciences Faculty Seminar.  We will enjoy a lecture from Prof. Dan Graur, the Moores Professor of Biology and Biochemistry at the University of Houston, as well as Professor Emeritus of Zoology at Tel Aviv University. His lecture will be titled:

The Emperor’s New Clothes: The Many Sins and Crimes of Biomedical Big-Data “Science”
(see abstract below)
Prof. Graur is an influential member of the Society of Molecular Biology and Evolution. He has made many constructive contributions to the field of molecular evolution, including studies on phylogeny, isochores, pseudogenes, and more. And yet, Prof. Graur is probably better known for his destructive role in the scientific community.

In this talk, Prof. Graur will demonstrate the role he has played repeatedly in calling the attention of the scientific community to malpractice of the scientific method, both within the field of molecular evolution and in the wider molecular biology research community. A beautiful example of this can be found in his masterpiece on chicken entrails and phylogenetic dating. In such cases, Prof. Graur typically acts as a single voice against a large group of highly driven and funded scientists. And yet, his exceptional ability to present the scientific rationale that was violated goes a long way to clear up misconceptions and common fallacies, and to right the wrongs brought about by flaws in the workings of the scientific community. One striking example followed the high impact avalanche of publications produced by the mega-consortium ENCODE in 2013, that pushed the headline “Death of junk DNA” all over both scientific and popular media. Prof. Graur was enraged and immediately set out to illustrate the fallacy in that claim using the molecular evolutionary perspective. His response to the ENCODE publications was titled “On the immortality of television sets“, and he embarked on an unprecedented campaign – a global tour under the banner “Kill ENCODE”. His upcoming talk in Haifa will generalize on this, as well as additional wrongs that afflict our modern scientific practice.

The Emperor’s New Clothes: The Many Sins and Crimes of Biomedical Big-Data “Science”
Prof. Dan Graur
Moores Professor of Biology and Biochemistry, University of Houston

Professor Emeritus of Zoology, Tel Aviv University

In biomedical inquiry, “data science” refers to studies that seek to extract meaningful insights from data sets that are claimed to be either too large or too complex to be dealt with traditional scientific methodology. Big data usually involves many cases (rows) and many traits (columns) and are thought to offer great statistical power. Sadly, big data also lends itself to unjustifiable but hard-to-detect manipulation.

In this lecture I will use examples from the literature—libel laws be damned—to discuss several problematic practices in big-data studies: (1) the abandonment of Popperian hypothesis testing in favor of 17th-century data-driven discovery, (2) the employment of methods and experimental setups that intentionally inflate or deflate the estimates of interest, (3) the employment of overly complex models leading to apophenia (i.e., seeing meaningful patterns in random data), (4) favoring discovery of true positives over the ability to distinguish true positives from false positives, (5) practicing the P-value cult, whereby statistical significance is confused with biological significance, (6) failing to recognize the existence of non-independent and overlapping categories, (7) employing arithmetical dirty tricks, (8) using “black boxes” prone to yield inferences unlikely to be applicable to any other data set, and (9) employing logical fallacies, such as “affirming the consequent.”

Some “crimes” related to “big data science” are due to external factors such as multi-authorship-distributed responsibility, ignorance of pre-big-science literature, and too much money.