BREDA, M. Navržení booleovské sítě na základě dat genové exprese u nemodelových organismů [online]. Brno: Vysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií. 2025.
The work of Maximilian Breda focuses on the inference of Boolean networks. In the theoretical part, he studied and described existing approaches to Boolean network inference. He then focused on gene expression measurement techniques and the processing of such data, with particular emphasis on RNA-Seq data provided in the form of preprocessed BAM files. In the practical part, the student designed an algorithm for network inference. I appreciate that the code was made openly available in a public repository. However, the evaluation metrics could have been described in more detail or reconsidered, and the choice of specific parameter values (as described in the Parameter Selection and Rationale section) would also benefit from clearer justification. Additionally, it would have been valuable to validate the results of the inferred network against existing knowledge of the bacterium’s regulatory pathways. Maximilian consulted his results regularly and consistently incorporated feedback. He worked independently and proactively. Overall, I consider all parts of the assignment to be fulfilled, although often in a rather minimalistic way. Therefore, I evaluate the thesis as good (“dobře”) and propose a grade of C / 75 points.
Student Maximilian Breda prepared his diploma thesis dealing with Boolean network inference in one selected non-model bacterium. The state-of-the-art part of the thesis is not elaborated well. This part is short, does not contain illustrations or mathematical equations that would be extremely helpful here. Particular topics are summarized without necessary details. For example, whole point 1) of the instructions “review of methods” is summarized in a single short paragraph mentioning only one method, point 2) “methods for gene expression” summarizes again only one method, particularly RNA-Seq. Therefore, instructions are met on a borderline threshold if at all. The practical part suffers from the same issues. Here, the absence of mathematical apparatus makes it impossible for reader to understand the proposed method. Moreover, github repository with code is not elaborated well, does not contain readme, license is missing and there is no tutorial to verify if it works. I do not understand, how “Parameter Selection and Rationale” is listed under the “Implementation” chapter. Here, the student states that “Five key parameters were carefully tuned to balance sensitivity against specificity”. This means that tuning was experimental part of the thesis and selection of final coefficients should be well supported by the results of such tuning! Unfortunately, also the description of the test dataset is missing, because student only briefly stated that three replicates (B, C and D) of C. beijerinckii NRRL B-598 were used. What are these? There is no chance for anybody to understand this. Results and discussion sections are of average quality. Evaluation of Boolean network inference is quite complicated but student defined a series of parameters which makes the whole part quite consistent and understandable. However, even here some issued can be found, i.e. Fig 3.1. is definitely not “in-degree distribution” but simply a network visualization. In-degree distribution would naturally be a histogram. Discussion is pretty well with minimum speculative statements, yet there are some (see my questions). Regarding the formal point of view, the thesis is terrible. Work with literature is awful and one cannot match particular pieces of information with references as those are placed after the whole paragraphs. The majority of tables and figures are not referred to in the text of the thesis. Tables 1.1. and 1.2 have captions instead of titles. There are periods missing, some sentences seem to be unfinished, “Boolean” is not always written with capital B, species names are not always written in italics etc. Since this is a selective double degree program and student should be awarded two diplomas, I don’t believe the thesis meets criteria posed by the study program to be defended.
eVSKP id 167537