Annex 7: Habilitation thesis reviewer’s report
Masaryk university
Faculty Faculty of Science MU
Habilitation field Biomolecular chemistry
Applicant RNDr. Radka Svobodová Vařeková, Ph.D.
Unit National Centre for Biomolecular Research, PřF MU, Brno
Habilitation thesis Analysis of Biomacromolecular Structural Fragments
Reviewer Doc. Mgr. Daniel Svozil, Ph.D.
Unit Laboratory of Informatics and Chemistry, Faculty of Chemical
Technology, University of Chemistry and Technology Prague
Reviewer’s report
Dr. Svobodová Vařeková works in the field of structural bioinformatics and cheminformatics.
Though her thesis is entitled ‚Analysis of Biomacromolecular Structural Fragments‘, it
actually covers a wide range of diverse topics only a fraction of which concerns
biomacromolecular fragments. The thesis consists of two parts: 50 pages of an introduction
and commentary and the collection of 16 peer-reviewed journal articles. The articles cover
four major topics: small molecule structure validation, the detection and characterization of
biomolecular channels, biomolecular structure comparison and partial atomic charge
calculation. Formally, the work is well designed, tables and figures are adequate and
informative and references are representative. However, some facts are described
ambiguously which makes reading and understanding the manuscript a bit harder. For
example, at page 4, methods for 3D comparison of organic molecules are categorized into
implicit and sequence alignment-based approaches. Nevertheless, this distinction applies to
atom pairing identification that is only a part of 3D comparison. In addition, though a term
‘organic molecules’ is used in the text, sequence alignment-based approaches can, probably,
be used only for biomacromolecules. Fortunately, these small slip-ups can’t conceal
undeniable scientific novelty Dr. Svobodová Vařeková contributed to each of her areas of
expertise, as is documented by papers from high-quality peer reviewed scientific journals. In
addition, I’d like to also point out to Dr. Svobodová Vařeková complex qualification. In her
habilitation thesis, she proved familiarity not only with many bioinformatics and
cheminformatics methods, but also with various approaches of computational chemistry. In
conclusion, the thesis clearly demonstrates Dr. Svobodová Vařeková ability to perform an
independent research using a sound scientific methodology.
Reviewer’s questions for habilitation thesis defense
1. At page 13, four different classes of algorithms (grid-based, sphere-filling, slice and
optimize, and Voronoi-based) for tunnel detection in biomacromolecules are briefly
described. What are advantages and disadvantages of these approaches? Why Voronoibased
method was chosen to be implemented in MOLE?
2. MOLE, CAVER and MolAxis are listed as three existing Voronoi-based methods. Two
(MOLE and CAVER) of these three approaches come from the same country, city and
university. What are the differences between latest versions of MOLE and CAVER? Are
there any plans to unite these two into one package?
3. Biomolecular structural comparison is one the most important tasks in structural biology
and bioinformatics. Thus, many different approaches were developed and published.
However, chapter 2.3.1., describing the state of the art in biomolecular structural
comparison, fails to give an overview of existing algorithms. What was the motivation for
the development of a new approach (SiteBinder) and how does SiteBinder compare to
already existing methods?
4. SiteBinder uses a combinatorial and subgraph matching approach to identify individual
atom pairings. Both approaches are computationally demanding, subgraph isomorphism
is, for example, NP-complete. What are complexities of these algorithms, as implemented
in SiteBinder, in Big O notation? Do there exist any alternative, faster methods?
5. OpenBabel is certainly an important cheminformatics toolbox. However, RDKit seems to
be at the cutting edge at the moment. The inclusion of EEM charges into RDKit, that
offers only Gesteiger charges, would certainly be beneficial for a wide cheminformatics
community. Are there any plans to do so and if not, I personally would urge you to do so.
6. The applicability of EEM charges in proteins was demonstrated on docking glycerol into
ubiquitin. In this experiment, EEM poses were compared to a QM pose that is considered
to be ideal. Why a QM pose, and not an experimentally derived structure of the glycerolubiquitin
complex (or any other ligand-protein complex), is taken as a benchmark?
7. Models for the prediction of pKa are derived only for uniprotic compounds and predicted
pKa is, thus, macrostate. How would you predict microstate pKa in multiprotic
compounds?
8. Approximately, how much faster are EEM charge calculations than QM charge
calculations?
Conclusion
The habilitation thesis ‘Analysis of Biomacromolecular Structural Fragments’ submitted by
Radka Svobodová Vařeková meets the requirements applicable to habilitation theses in the
field of Biomolecular chemistry.
Prague, 12.8. 2016
Doc. Daniel Svozil, Ph.D.