Too little science in my new "supervisor" group

Since October 2012 the administration of Swedish University of Agricultural Sciences, SLU is forcing me to accept a new group of supervisors. This group is not legitimate, because it was appointed in conflict with all prescribed rules. Anyway, new supervisors constantly gather for meetings and produce protocols. Reading these protocols makes me feel sorry for SLU administration: it so difficult to find smart people for a hatchet job. A true scientist would not take this job, and a fake scientist lacks accuracy and diligence to do it well. One of the first protocols that I got from the new “supervisor” group demonstrated it very clear.

At half-time my PhD project has “received praise from the external evaluator for the choice of fundamental biological questions to be investigated, as well as for the elegant systems which has been developed”. It was also obvious that obtained material would be more than enough for a PhD project. My new “supervisors” could not argue with this, instead, they said that the method that I used is outdated. They argued that it might be a problem to publish the results, and my work cannot be recommended for a PhD degree. Unfortunately for them, the statistics of the recently published articles shows that the method is not outdated (see comments below). Plus I did not have any alternative, because there were no other methods available at the Department at the time when I designed experiments and later I was never offered the opportunity to use new and expensive methods. When my new “supervisor” was caught in this attempt to discredit my research, he immediately stepped back, saying that he was misunderstood. Later, he was trying to pretend, that it was just a “scientific discussion”.

Below are my comments on a protocol from supervisor meeting and a response from a “principal supervisor” to these comments:

Why put DGGE method to death?

Comments on protocol from supervisor meeting for Elena K. 2012-12-03

I stated before and continue to insist that I nether recognize nor accept the supervisor group appointed for me by SLU administration in violation of the prescribed procedure. Particularly, I strongly oppose Mr. H and Ms. C candidacies, because I consider them biased toward my research.

Nevertheless, I would like to comment the protocol from supervisor meeting from 2012-12-03, signed by Mr. H and Ms. C, because the authors provided me with an opportunity to demonstrate their preconceptions.

Part I. DGGE is an adequate and not outdated method for microbial ecology

In the protocol from Mr. H and Ms. C there was no discussion about the scientific questions and the results of my projects. The main point of the comments was that DGGE method, used in my studies, is outdated . To make the discussion understandable for scientists from other fields of biology, a short introduction of our routine is necessary.

The subjects of studies in microbial ecology are microbial communities. Native microbial communities consist of many microbial species. Because we do not have detection systems that allow us to study microbial communities directly, the first step in each experiment is to amplify the signals from every group of microbes. This is done with polymerase chain reaction (PCR). So, PCR is simply the method to increase the signal up to detectable level.

The second step is to detect the amplified signal. For this there exist several methods. At our Department cloning-sequencing, T-RFLP, DGGE and NGS were/are used. Since, first two methods are already abandoned by our Department, I will compare the last two.

I. DGGE is about 100 times cheaper than NGS

Table 1. Cost of detection methods

Method of detection	abbreviation	Cost for 96 samples (plate)
Denaturing Gradient Gel Electrophoresis	DGGE	~750 SEK
Next Generation Sequensing	NGS	~100 000 SEK

2. DGGE is the only detection method that is left in the Department

DGGE is performed in home. NGS means buying a service. Next Generation Sequencing in spite that it sounds very impressive, on practice means that PCR is done at home and then products are packed into envelopes and sent to a company, which actually does sequencing. Results come as a huge dataset and processed with algorithm on an autopilot mode.

SLU is supposed to teach students and provide environment for them to master methods and techniques. It should be both, basic and advanced methods, because:

1. advanced methods does not always substitute basic ones
2. there are many students from the countries, where expensive techniques are not available

So, the main task of teaching unit should be preservation and improvement of basic methods and accommodation of new techniques that can be performed in home.

In spite of this, currently at our Department students can learn only DNA extraction and PCR. In seven years that I spent at the Department I could not master any new technology, which would make me more competitive at the work market after PhD. Even to study DGGE I went to the other laboratory, because there was no expert at home. No need to say that a student, in order to master a new method, needs supervision of an expert and a possibility to consult one on every day basis. Short-term visits to other labs do not provide necessary conditions to learn new methods.

In spite that DGGE currently stays the only one detection method available in home, there were constant and strong efforts to put it to death. The possible reasons for killing DGGE method are provided in the next paragraphs.

3. DGGE leaves no space for fraud and data manipulation

DGGE, as a method of detection, is absolutely intolerable to bad quality of PCR. If investigator gets bad PCR it is visible on a gel (examples are provided below). So, DGGE demands high accuracy of previous steps and leaves no space for fraud, since reviewers can easily request for original pictures of the gels.

In case of NGS, samples sent for sequencing do not pass proper quality control. Later, results arrive from company in electronic format (google can give examples of data from NGS on request). If there is no strict control for PCR quality BEFORE sending samples for sequencing, then there is a lot of possibility for incorrect results.

To illustrate this, I would like to show some results from my experiments.

Why there is a need for quality control?

Before starting to study natural communities I tested how the approach works on artificial simple community that consist of 4 bacterial species. So, if there are 4 species in the test tube, should be 4 different PCR signals and 4 bands on the gel (fig. 1)

However, polymerases (enzymes that perform PCR) that are in routine use at our Department were unable to amplify mixed template correctly (fig. 2A and 2b). Although there were only 4 bacterial species in a test tube, outcome results showed complex communities consisted of more then 20-50 species! So, being suitable for other purposes these polymerases cannot be used to study communities. After testing several options I found BioTaq polymerase that (in combination with certain primers) was suitable for routine use (fig 2 c).

Fig 2. What can happen when unsuitable polymerase is used for community analysis!

(Remember that it should be only 4 bands in mixed community and a single band for each individual species!)

1-3, 8-10, 15-17, 22-24, 29-31 – mixed community (3 replications); 4-7, 11-14, 18-21, 25-28, 32-36 – individual species A,B,C and D

(To whom it may concern) primer sets that were tested 1-7 – Muy1f-Muy2r; 8-14 – Eb246-1406R; 15-21 - EubB(27F)-U926R; 22-28 – EubB(27F)-U1510; 29-35 – BrunkBac 41-61-EubA(1522R)

a. DreamTaq polymerase

b. Phusion polymerase

100. 
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     с. BioTaq polymerase (in positions 22-28 PCR correctly reflected what was in test tubes)
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     So, for bacterial communities I managed to find suitable conditions, but...
     
     
     There is no way to study fungal communities with existing approaches!
     
     
     With fungal communities situation is worse. I did not find any PCR conditions that correctly reflected composition of artificial community consisted of 4 fungal species. It is not even possible to get only one band from single fungal species (fig 3). The prerequisite for community analysis is that each species gives single distinct signal. However, no difficulties were reported from other researchers from the department, who use NGS as a detection method.
     At least it remains very questionable approach to send samples for expensive sequencing without testing for PCR quality, and DGGE is a method of choice for such test.
     
     
     Fig 3. Fungal community never went fine!
     Amplification of artificial community consisted of 4 fungal species (named A,B, C, and D) under different conditions (the best outcome I managed to get)
     Mix A B C D Mix A B C D Mix A B C D Mix A B C D Mix A B C D Mix A B C D Mix A B C D
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     
     If PCR would work correctly then DGGE would be method of choice for studying fungal communities, since DGGE can easily resolve 60-70 bands and fungal community rarely exceeds this level of complexity (especially, when DNA is extracted from small (0.5 g) amount of soil). The main problem for studying fungal communities is inability of existing approaches to provide correct signal amplification from native communities. Before this problem would be solved none of the detection systems can compensate this drawback. However, NGS can easily hide it.
     
     
     
     
     3. DGGE is not that abandoned as it was presented!
     
     
     I never judge method by how fancy it is, but for modish persons there is some information below.
     Search in Scopus gave 842 hits for “denaturing gradient gel electrophoresis” published in 2012 with impact factor of journals (checked just on the first page) being:
     Clinical microbiological review 16.129
     ISME journal 7.375
     Bioresource technology 5.352
     PLoS ONE 4.351
     International Journal of Food Microbiology 3.847
     Environmental Microbiology 3.829
     FEMS Microbiology Ecology 3.408
     Molecular Ecology Resources 3.062
     
     
     
     
     
     
     
     
     
     
     Conclusions:
     1. DGGE is 100 times less expensive detection method that NGS.
     2. It can be done in home. It can be used at least to check the quality of samples before proceeding with expensive analysis
     3. The high resolution that NGS provides is not necessary for analysis of fungal communities, but this method provides good cover for errors.
        
        
        
        
     
     
     Part II. Advisers forgot what they said earlier, but “we have all moves written down”
     
     
     In their protocol Mr. H and Ms. C said:
     
     
     
     
     
     When I started my experiments in 2008, there was only two method of detection in use at the Department: T-RFLP and DGGE. So, counting on what there was available I designed experiments for which knowing names of each individual microbial species was not necessary for answering the questions. Nevertheless, I was ready to master any new method. However, when NGS became available at the Department I have never ever been offered an opportunity to use it.

     At my half time I have been recommended to cut bands form DGG gel, extract again DNA from each band, do PCR for each individual band separately and send product for sequencing to find out the name of the species. Here is a document:

     
     

     

     
     

Each band on DGGE is 0.1-0.3 mm thick (like a hair) and 3-4 mm wide. The distance between two adjacent bands is 0.3-3 mm. One sample produced from 30 to 60 bands and one experiment contains 124 samples to analyze. In addition these gels are very prone to cracking. If I should follow the advice given by Mr. H on my half-time seminar, all together I need to cut (doing it wearing a protection and under UV light!) and analyze from 3720 to 7440 bands . When I pointed out that such work would take at least a year, I was given an advice to cut at least “some of the bands”. On my question what should be criteria to select bands for analysis there was no answer provided...*

_______________________________________________

*Once upon a time rabbits became very tired of being bulled. They came for advice to old and wise owl and said: “O, owl! Please, teach us what we should do! We are so tired of being bulled by everyone in the forest!” Owl thought for a while and said:” Rabbits, you need to become hedgehogs!” Rabbits jumped away very happy, but after a while came back and asked owl again: “Owl, but HOW can we became hedgehogs?” Owl looked at them and said:” How should I know? I am not a tactician, I am a STRATEGIST!!!”

Part III. Even IF I would be able to follow Mr. H advice, does it worth to do?

After silver staining PCR is impossible. Other stains do not provide good resolution.

Gels after DGGE analysis are stained with silver. After silver staining DNA is not suitable for PCR (faint bands, fully developed), so it cannot be amplified for sequencing. SYBR green can be used to stain gel in order to be able to use DNA from band as template for PCR, but SYBR green is at least 5-10 times less sensitive and thus, reduce the value of analysis, which I need to perform to answer the main questions of the experiments.

Aiming to identify microorganism only by part of one gene – is overambitious and not scientific.

Even complete 16S rDNA or ITS region sequences are not enough for species identification. And identification is the first step BEFORE microorganism can be correctly classified. Sequences that are used for microbial community studies are far too short to provide reliable information for organism identification. Serious scientist discussed whether it is possible to identify species correctly even knowing sequences of several genes. Thus, sequencing alone does not provide correct information on taxonomy of microorganisms and therefor is useless.

More, Nesseria meningitidis and N. gonorrheae on the basic of their genomic and even metabolic properties would have been considered conspecific, however everyone knows the differences...

Do we have control of how many sequences are only the results of a polymerase mistakes?

Polymerases that are routinely used for community analysis usually do not have proof-reading activity. It means they do considerable number of mistakes when copy original sequence from bacteria or fungi. In manuals, the rate of mistakes that polymerase does is given for optimal conditions, however, environmental samples:

1. have damaged DNA
2. contain many impurities
3. often need PCR enhancers

And all these factors can alter polymerase behavior and increase the rate of mistakes. This is one of the possible causes for increase in the number of bands, compared with expected number after amplification with DreamTaq polymerase (fig. 2a).

Even if the name of microbe would be found and even if it would be correct how much information it gives about function?

The same microbial species can perform different functions in different communities, or under different environmental conditions (temperature, pH, availability of nutrients, etc).

This list can be continued...

Part IV. Conclusions.

Concluding all above mentioned, I state that:

1. My experiments were designed according the tools that were at my disposal at that time
2. I have never been offered an opportunity to apply NGS (if I would be offered now I will do it!)
3. Recommendations given to me by Mr. H and Ms. C I consider as non-scientific approach and not worth doing

Uppsala, January 7^th, 2013 

 Elena K

________________________________________________________

From: Nils H

Sent: 07 January 2013 23:01

To: Elena K; Marianne C; Martin W; Johan M; Pär F; Christer B

Cc: Jan S

Subject: Re: why put a method to death+ next supervisor meeting

Dear Elena,

You are right that the research questions are far more important compared

to methodology. In order to complete your thesis and for possible

discussions with the supervisor group you have to complete the study plan**

which was due to the faculty already Nov 15 2012. It is not possible to

have such an in depth discussion without a study plan as a basis. In the

supervisor group, we are prepared to have such discussions about your work

provided we have a study plan sent to us as a starting point for the

discussion. Furthermore, the work will not be fruitful if you are not

present yourself at the meetings.

The time for the next supervisor meeting is 13.00 Jan 10 in room A372 as

announced before.

Sincerely,

Nils H

** According the new "supervisor" group the information that my updated research plan lacked was: