More Bad Science in Predatory OA Journals

Cosmotherm

The suspect article.

The COSMO-RS method is an advanced method for the quantitative calculation of solvation mixture thermodynamics based on quantum chemistry. It was developed by Andreas Klamt and is distributed as the software COSMOtherm by his company COSMOlogic (as well as in the form of several remakes by others).

Some Nigerian researchers have used the software (without a license) and report a tremendously and completely unbelievably good correlation (r²=0.992) between the predicted results and experimental data for the logKow (octanol water partition coefficient) of ionic liquids.

A closer look to the data reveals that the experimental data in the plot have nothing to do with the data reported in the PhD thesis referenced as the experimental data source. Hence we need to conclude that the entire paper is complete fake.

The unrealistic results are published in the Journal of Environmental Chemistry and Ecotoxicology, a journal published in Faisalabad, Pakistan by AcademicJournals, one of the very first publishers included on my list of predatory publishers back in 2010.

I received an email from Prof. Dr. Andreas Klamt in Leverkusen, Germany who explained the case to me. He’s the CEO of COSMOlogic. Dr. Klamt writes:

I had been told there about a case, where some Nigerian people reported unbelievably good results with my COSMO-RS method. They report an r² of 0.992, completely out of the realistic expectations. Checking in more detail I found that the experimental data have nothing to do with the experimental data reported in the PhD thesis which is given as source of the experimental data. Not even the number of experimental data is in agreement with the source, nor the sign of the logarithm. The data are complete fake.

A second published article written by the same authors and appearing in the same issue of the same journal is “Improvement of carbon dioxide absorption technology using conductor-like screening model for real solvents (COSMO-RS) method.

Improvement of carbon dioxide absorption technology

A second article in the same issue.

For both articles, all the authors show their affiliation as the Department of Chemical Engineering, Ahmadu Bello University in Zaria, Nigeria.  The two articles have two authors in common: S.I. Mustapha and S.M. (Saidu Mohammad) Waziri. Mustapha’s name does not appear on the department’s faculty list.

It’s not uncommon to see more than one paper by the same author or authors in the same issue of a predatory journal. Predatory journals sometimes grant volume discounts to authors.

acetaldehyde

acetaldehyde

Dr. Klamt has tried in vain to contact the authors, the journal’s editors, and the publisher, with one exception. He reports that one editorial board member responded saying he has been trying to get his name removed from the journal’s website for a long time, but without success.

Predatory publishers accept papers regardless of their scientific rigor just to earn the article processing charges. Junk science like these articles is polluting and corrupting the scholarly record.

15 Responses to More Bad Science in Predatory OA Journals

  1. Samir Hachani says:

    Well it seems that certain regions of the world tend to have a high ratio of these ” scientists “.I like Dr. Beall’s ” Predatory journals sometimes grant volume discounts to authors “.It sounds like a supermarket…….

  2. Farid says:

    Dear Dr Beall
    You made a valuable comment about so called predatory journals. However, you always blame OA publishers for such behavior and this activity is not limited to OA ones, Here is a sample from a well known journal, http://www.michaeleisen.org/blog/?p=1439
    Farid

  3. tekija says:

    Science??

    Incidentally, an e-mail today shows great innovation, or should I say, imagination, in advertising the “great” open access movement (with an 80% discount, of course) on behalf of the “British” Journal of Medicine and Medical Research!

    http://www.sciencedomain.org/page.php?id=authors-speak
    http://www.sciencedomain.org/page.php?id=author-profiles

  4. wkdawson says:

    A couple of minor points.

    + It is a common custom these days to have the PI listed as the last person on the author list. Therefore, the first author may be a student.

    + As I commented previously on the Science sting operation, all journals (preditory or not) depend on submissions that are done in good faith.

    The fiasco in the physics world with Schon ( http://www.americanscientist.org/bookshelf/pub/physics-and-pixie-dust ) is an example where Science and Nature were both duped.

    It probably makes sense for OA journal to be extra careful. You know generally what comes out of Science (though they did publish 8 of Schon’s bogus research articles, even without a discount). OA, on the other hand, is likely to have both gold and dross mixed together. If OA takes the publishing seriously, they have be serious about handling cases of fraud.

    Just glancing at one of the papers, the fraud isn’t immediately obvious, though the fits do appear to be a bit too good. It also seems like there are not enough details to know how to reproduce the results. It would be doggedly quite time consuming for any reviewer (who is not also a competitor doing exactly the same research) to try to reproduce several of the results, and even if the reviewer tried, there are all sorts of settings in these programs that could make the calculations come out different. I don’t even think Nature and Science expect the reviewers to spend several days of their time evaluating a manuscript. So, even for good journals (OA or not), this whole system must rely on good faith, both on the part of the authors and on the part of the reviewers. The system breaks down without it.

    Therefore, it seems like it is not so much about the fact that some journal (any journal) has published something that is fraudulent; rather, it is what that journal does about it when that fraud is exposed. There we may identify the difference between an OA journal that is serious about publishing scholarly research and one that is not.

    • Merry says:

      I thought the point of a reviewer WAS to spend several days evaluating a manuscript. Certainly the researchers I’ve worked with have always taken their reviewing very seriously.

      • wkdawson says:

        I certainly do, and I spend that kind of time, in fact.

        However, that does not seem to be how most peer reviewers have dealt with the job as far as I can tell. At any rate, that still ignores that fact that submissions should be done in good faith. If our job is to detect forgeries, we need a lot more time than two or three days to reproduce the results, etc.

      • wkdawson says:

        It would be interesting to know how much time the reviewers spent on the example that Farid cited (see link above). That was a review process done on a submission to Science.

        Was that _really_ the conclusion of a serious three day review from at least two reveiwers?

        There was one review I did for one of the Nature group where I could see that the other reviewers took the job very seriously. Some of the people who have reviewed my work definitely took some serious time out to read it and think about it. However, both OA and non-OA, and even in high impact journals non-OA journals, I’ve witnessed my share of some very slipshod and unethical behavior on the part of reviewers.

        The system is definitely flawed. Lots of washing of hands: perhaps all the way up the ladder. This lack of good faith makes authors cagey. It also renders an editor’s job difficult because a good faith exchange between reviewer and author becomes impossible. That means even the best editor with the best intentions delivers a mediocre to poor assessment. In the midst of this, there are the people who submit fraudulent work that preys on this lust for grandeur (high impact factors, news blurbs, prestige, etc.).

        I don’t have any idea how to fix it other than to say that everyone has to be ethical in this process. Without good faith, the entire edifice pulverized to dust.

  5. A. Meyer says:

    If the software is not open access and the methods used are confidential secrets of the software sellers to begin with, how can anyone double check or reproduce the results ?

    Any results desired can be obtained with a cleverly programmed black box.

    • wkdawson says:

      There are some freely available molecular orbital calculation programs that have some version of COSMO included, though they may not have all the features. ORCA is one that I know of and use for some of my work, but there are others that at least have some means of access within academia for a professor (at least).

      Although it is good for one who does this stuff to at least take a crack at writing a simple molecular orbital program to appreciate the difficulty of the job, as a generalized application, it is a very tough type of program to write and maintain. Some people really do have to specialize in building and maintaining these projects.

      Typically, people who do molecular orbital calculations have access to the Gaussian program where g09 has an option for cosmo-rs. (Gaussian was largely John Pople’s lifetime project, which is now carried on by others.) So there would be ways to check the purported results of the manuscript. However, molecular orbital calculations can cost days to run on parallel computers. So it is not always cheap to check this kind of stuff.

      Another problem that is somewhat close to what you allude to involves settings and how the program is used. COSMO-RS is an implicit solvent program. To put it in layman’s terms, we have a molecule and we surround it by an amorphous blob of jelly with certain conducting properties (COSMO: conductor like screening model), and hope that this resembles an actual solvent that would surround this molecule. Water does have different properties than hexane for example. Again, because of extreme cost in time for calculating molecules in general, a program like COSMO saves a lot time if and only if the problem is one where the configurations of the solvent can be assumed to be smeared out so that, on average, they appear to resemble this amorphous condition. These programs all have their share of instabilities and so one finds with real problems that parameters have to be fiddled with. The biggest issues are when to use these _approximations_ and how to use these _approximations_ properly, and that is not something that is always obvious.

      So, you have a point, but keep in mind that there is considerable nuance in how this plays out in the real world of (honest) battling with real research questions.

  6. A. Meyer says:

    COSMORS is a proprietary commercial black box .

    You have no way to replicate its results independently, check for programming mistakes or even be authorised to use your own results to attempt to find out the values of the parameters programmed (read the license).

    Any and all publications using COSMORS are based on faith that the black box does what the software vendors claim. This is not verifiable or scientifically sound either.

    • wkdawson says:

      I can mostly agree with you. I don’t particularly like implicit solvent models in general: not only because of the black box character of them, but also because the physics of such mean field methods tends to ignore important aspects of solvent/solute interactions (particularly hydrogen bonding). That is another really negative point about the papers in question here; even if the study had been done “correctly”.

  7. Kaltenbrunner says:

    Indeed the COSMORS model was conceived for neutral molecules and inherited all the limitations of implicit solvent models. It should not be used for Ionic liquids or working with hydrogen bonds.

    This experimental paper clearly shows COSMORS fails for non neutral molecules with hydrogen bonds,

    Evaluating the COSMO-RS Method for Modeling Hydrogen Bonding in Solution – Sofja Tshepelevitsh, Merit Oss, Astrid Pung, and Ivo Leito

    “When the HB acceptor is an anion, the correlations are poor and in some cases even qualitative predictions fail.”

    These papers about ionic liquids shows COSMORS fails to describe the non-specific effects in ionic liquids

    Computational studies on organic reactivity in ionic liquids
    Phys. Chem. Chem. Phys.,2013, 15, 412
    Cinzia Chiappe and Christian Silvio Pomelli*

    Reply to the comment on ‘‘Computational studies on organic reactivity in ionic liquids’’ by C. Chiappe and C. S. Pomelli, Phys. Chem. Chem. Phys., 2013, 15, 412
    Christian Silvio Pomelli* and Cinzia Chiappe

    As indicated above there are many papers published using COSMORS whose authors believe the vendors marketing hype without understanding the intrinsic limitations of the model.

  8. Michael says:

    Dears,
    not only such bogus articles spoil the hard work of real scientists but they also shed – wrongfully – a bad light on the methods and software they claim to have used.

    Meyer, not everything in life is free. Commercial software is not. But this does not mean that it is black box, which COSMO-RS is definitely not, nor it does mean that you cannot reproduce it’s results. But it requires an effort to study the relevant literature on it, Meyer.

  9. Sven Müller says:

    Michael

    Here is a comment “aus erster Hand” or “straight from the horse’s mouth”

    http://ftp.ccl.net/chemistry/resources/messages/2009/10/23.004-dir/index.html

    “Admittedly, in our COSMOtherm implementation of COSMO-RS we have a few parameters being not completely disclosed.”

    Do you or anyone but the vendor using this software can recreate its results ?

    If so please publish the “few parameters being not completely disclosed” AND their values so other people can believe in the results without resorting to faith. Faith is not part of any sound scientific methodology.

    How many parameters are “few parameters being not completely disclosed” ? 3 or 30, or 300 ….

    It is also possible to refit on the fly all these parameters using the built in database of values and always provide a better answer than the other remakes of a method that is not fully disclosed, this is what is meant by black-box software, one where the user can’t really retrace how it computes. It is secret and proprietary, why should anyone blindly believe in those results ?

    BTW If you try to find Klamt at the Uni to ask these questions you will have a hard time because he is not listed in their directory. After reading comments as those posted online I decided not to waste any time on this methodology.

    I wish you good luck reading the literature., Michael

  10. […] experimental data for the logKow (octanol water partition coefficient) of ionic liquids. Source : https://scholarlyoa.com/2013/10/24/more-bad-science-in-predatory-oa-journals/ 37 38. Penalty for Plagiarism Outside of academia the problem of plagiarism continues to generate […]

Leave a Reply -- All comments are subject to moderation, including removal.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: