Gilderoy Lockhart produced a new PhD!

Recently my former supervisor baked a new PhD. I had a hope that after getting in troubles with me, he might revise his supervisory practice. Unfortunately, it doesn't happened and Prof.  Lockhart continues to release PhDs in the same way as before.

There are too many things in the thesis of the new PhD candidate that can be questioned. I will discuss below only two points that made me to grab my head in disbelieve of what I read. The thesis title is 'Root associated microbial communities ofdifferent strawberry cultivars as influenced by soil type,verticilium, and biofumigation.' The 'discovery' that shocked me the most was the enormous numbers of fungi that were found in the soil. For the analysis Prof.  Lockhart and his student used 0.5 g of soil. (see the picture below: this is how 0.5 g of soil looks like) 

Now quotation from the thesis: “Bioinformatic analysis of rhizosphere soil revealed 16923 pyrosequencing reads that passed the quality control checks revealed 589 clusters, 86% of which were of fungal origin.” This means that 589 different types of DNA molecules were found, 86% of which are fungal DNA. This gives us 506 different types of fungi living in 0.5 g of soil. If it would be so the “soil” would look like a ravel of fungi without any real soil particles. It is similar to statement that herds of antelopes can be grazing in the kitchen. Nevertheless, the discovery was not questioned and Prof.  Lockhart went on to analyze the data.

Where this striking number of fungi might come from? There are at least two sources: contaminated lab and inappropriate protocols that always successfully provide researchers in my department with a level of diversity necessary for impressively-looking publications. But these issues will be discussed some time later in another post.

The unnatural number of fungi discovered in a pinch of soil was already enough to doubt the results, but the freakish data analysis made things even worse. The statistics based on presence – absence data showed no differences between treatments. To overcome this obstacle the quantitative analysis was performed. It means that bands (produced with erroneous protocol) were measured for their fleshiness. If one band was thicker then the other it was concluded that one microbe presented in the sample in a bigger amount then the other one. It called “relative abundance”. 

 Let's see an example. Assume that we have two samples. In sample A there is 1 microbe and in sample B - 100 microbes. Such small amounts are beyond the detection and we need to amplify the signal. Now imagine that during analysis the signal from sample A was amplified for 100000 times and from sample B only for 10 times. Then the relative abundance in sample A would be 100 times higher then in sample B. From this result the erroneous conclusion can be done, that in the sample A there was 100 times more microbes then in sample B. 

This kind of error happened very often! That is why the long list of “must-to-do” things was developed for quantitative analysis. To monitor the correct signal amplification different types of controls must be used. Plus, the analysis for each sample must be repeated several times to confirm that every time the same results are obtained. Plus the calibration must be done to set a scale.  In this work, however, even the most basic demands were ignored. Casting Relative Exuberanti spell Prof. Lockhart and his student conjured all necessary differences and spectacularly culminated thesis with 19-points conclusions!

Who is guiltier in doing such kind of work, student or professor? I think that 99% is the supervisor fault. Student usually spends considerable amount of time before he discovers that the research is predisposed to be a fraud. When suspicions grow into certainty it's already too late: too little time left, no other methods available in the department and nobody around who does research in a different way. So, the routine is reproduced year by year and anyone who dares to question ill practice is outcast.

It's difficult to calculate a total cost of such studies; but only the salary of one professor (about 55.000 SEK per month) and one PhD student (av. 22000 SEK per month) cost taxpayers about 1 000 000 SEK per year.