IMR Press / JIN / Volume 20 / Issue 2 / DOI: 10.31083/j.jin2002028
Open Access Original Research
The connection between rs6265 polymorphism in the BDNF gene and successful mastering of the video-oculographic interface
Show Less
1 Laboratory of Medical Cybernetics, Voronezh State University, 394018 Voronezh, Russia
2 V. A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences, 117997 Moscow, Russia
3 Department of Genetics, Cytology and Bioengineering, Voronezh State University, 394018 Voronezh, Russia
4 Laboratory of Metagenomics and Food Biotechnology, Voronezh State University of Engineering Technologies, 394036 Voronezh, Russia
5 Department of Programming and Information Technologies, Voronezh State University, 394018 Voronezh, Russia
*Correspondence: (Artem P. Gureev)
J. Integr. Neurosci. 2021, 20(2), 287–296;
Submitted: 27 January 2021 | Revised: 5 March 2021 | Accepted: 20 April 2021 | Published: 30 June 2021
Copyright: © 2021 The Author(s). Published by IMR Press.
This is an open access article under the CC BY 4.0 license (

A video-oculographic interface is a system for controlling objects using eye movements. The video-oculographic interface differs from other brain-computer interfaces regarding its improved accuracy, simplicity, and ergonomics. Despite these advantages, all users are not equally successful in mastering these various devices. It has been suggested that the genetic characteristics of the operators may determine the efficiency of video-oculographic interface mastery. We recruited healthy users with rs6313, rs2030324, rs429358, rs10119, rs457062, rs4290270, and rs6265 polymorphisms and analyzed the relationships between these polymorphisms and values of success in video-oculographic interface mastery. We found that carriers of the G/G genotype of the rs6265 polymorphism (BDNF gene) demonstrated the best results in video-oculographic interface mastery. In contrast, carriers of the A/A genotype were characterized by large standard deviations in the average amplitude of eye movement and the range of eye movement negatively correlated with goal achievement. This can be explained through the fact that carriers of the A/A genotype demonstrate lower synaptic plasticity due to reduced expression of BDNF when compared to carriers of the G/G genotype. These results expand our understanding of the genetic predictors of successful video-oculographic interface management, which will help to optimize device management training for equipment operators and people with disabilities.

Video-oculographic interface
Brain-derived neurotrophic factor
Eye movement
Working memory
1. Introduction

Scientific and technological signs of progress in computer technology have made a wide range of digital control systems available to users. First, this applies to systems designed for mass use, which have a relatively low cost and do not require significant time for user training. These include input devices, such as keyboards, and non-keyboard input devices, such as mouses, joysticks, touchpads, touchscreens, and trackballs [1]. However, these devices are not always suitable for people with disabilities [2]. Therefore, researchers are actively developing a class of devices based on alternative finger activity mechanisms for generating user commands. An example of this is the so-called “alternative human-computer interfaces”, in which the user generates commands through mechanisms such as electrical brain activity (brain-computer interface) [3], electrical muscle activity (myoelectric hand prostheses) [4], changes in the position and activity of body parts (accelerometric interface) [5], or movements of the eyes (video-oculographic interface) [6]. In this respect, video-oculographic interfaces (VOIs) are of the most significant interest due to their high accuracy, simplicity, and ergonomics. First, however, it is crucial to search for correlates associated with compelling performance by users of this class of devices.

Interest has increased in understanding the relationship between users’ cognitive state and eye movement features during human-computer interfaces. However, a deeper understanding of this relationship is still required [7]. In addition, there is a significant gap in the genetic determinants of the success of human-computer interfaces operation. This is likely since conducting such research requires integrating specialists in neurophysiology, digital signal processing and molecular genetics. There is some research of relationships between brain-computer interfaces and “genetics”, but under the “genetics” term are presented various variants of mathematical algorithms for signal processing [8].

Our preliminary results have demonstrated connections between the number of single nucleotide polymorphisms (SNPs) and values of systemic behavior during device operation (accuracy of achieving goals, characteristics of subjective time scales) [9, 10], as well as characteristics of biomedical signals, such heart rate variability [11]. Previous studies have shown that SNPs in a gene encoding a protein involved in serotonin metabolism are associated with the mastery of the electromyographic interface [9]. Of particular interest are polymorphism in the brain-derived neurotrophic factor (BDNF) gene (such as rs6265 and rs2030324), which is responsible for the regulation of the maturation and differentiation of neurons and forming of working memory [12] and long-term potentiation [13]. In addition, 5-hydroxytryptamine receptor 2A (HTR2A) (rs6313 polymorphism) and tryptophan hydroxylase 2 (TPH2) (rs4570625 and rs4290270 polymorphisms) genes are involved in the serotonin synthesis and metabolism, which play a critical role in human cognitive performance [14]. The 20 Kb region containing both the translocase of mitochondrial membrane 40 (TOMM40) and apolipoprotein E (APOE) genes unique thus far in complex human diseases (including cognition impairment such as Alzheimer’s disease), which is statistical support [15]. rs10119 and rs429358 polymorphisms were studied from this region. This research focuses on different genetic factors that affect the brain -interfaces technology, in so far as assessing correlations between VOI mastery and SNPs in a gene involved in regulation of serotonin metabolism and cognitive characteristics in humans.

2. Material and methods
2.1 Experimental participants

40 volunteers took part in the experiment, comprising 22 males and 18 females aged between 19 and 23 years; the median age was 20. The volunteers all had no neurological or psychiatric pathologies, did not take drugs affecting their movement coordination or decision times, and had normal or corrected normal vision. Ethics Committee approved the current research of Voronezh State University. All experimental participants gave informed consent for an EEG in accordance with the Helsinki Declaration. Informed written consent for enrollment in the research was obtained from all the participants, who were informed of any subsequent genetic testing conducted using their biological material. The experimental design was presented in Fig. 1.

Fig. 1.

Experimental design.

2.2 Video-oculographic interface

Software previously developed by the authors [16] was used in these experiments. The software simulates controlling the movement of a virtual marker in 2D space and, thereby, provides training and testing of a user’s VOI mastery. The software integrates the hardware of the VOI with a computer, thus allowing the experimenter to set the parameters for video recording of the movement of the virtual marker, calibrate the user’s eye movements, and record the user passing through a series of tasks formed by the experimenter. Via eye movements, the user moves the virtual marker across the field from one area to another while bypassing obstacles, the position of which can be changed by the experimenter. Eye position coordinates, marker coordinates, and collisions with obstacles were recorded (Fig. 2).

Fig. 2.

Example of the virtual marker movement in 2D space during task performance in the VOI. Space included start and goal point and four obstacles (black box). The user’s task is to move the virtual marker via eye movements from the start to the goal point bypassing obstacles. The alignment of a marker with an obstacle was regarded as an “error”.

Users were positioned in front of the monitor screen (22 inches) at a distance they found comfortable. With the VOI, it was possible to obtain a sharper image using a standard video camera with an extracted infrared filter. In addition, a holder was made for the infrared camera that was attached to the head and provided the ability to change camera position. This made it possible to adjust the image coming from the camera, regardless of the head’s shape, in order to meet user convenience requirements [6].

After calibration, an experimental field was launched on the monitor screen, including the launch area, rectangular boundaries of several objects, and the target area. The instructions presented to each user were explained to attempt to avoid combining the marker with the obstacles (objects indicated as black rectangles) on the screen. The alignment of a marker with an obstacle was regarded as an “error”, which was automatically recorded by the program. The program recorded and recognized movement of the pupil in four directions: “up”, “down”, “left”, and “right”. A new command could be generated by the user no more than once per second. Before each experiment, the necessary device calibration was performed. Each user was given four trials to complete the experiment, each lasting two minutes. The time between trials was from three to five minutes. The first trial was training and not taken into account in statistical analysis. Second, third and fourth trials were measured. The total number of measurements was 120 (three trials for 40 participants).

2.3 Assessment of eye movement parameters

The Haar wavelet discrete transform algorithm was used to assess the parameters of eye movement [17]. First, the entire time series’s mean values and standard deviations of the entire time series were calculated for the obtained values and each level of the wavelet transform (Fig. 3). Analysis of the mean values of the coefficients of wavelet transforms—shows the average level of representation in the signal of this wavelet, and standard deviation shows the level of spread of the signal variability. This is important since the eye movement is not constant but salutatory. Therefore, the obtained values are interpreted as the average distance traveled by the users during the experiment. In contrast, the change in distance corresponds to the wavelet transform scale (or level). Accordingly, the standard deviation of the time series data after the wavelet transformation is the average values of the pupil displacements from the average displacement amplitude for the entire time of the test. The obtained values were called approximating sequences (a-sequences) and detailing sequences (b-sequences) in accordance with the approaches adopted in wavelet analysis. In this case, the wavelet transformation was carried out directly on the data, where the sampling frequency was artificially reduced to 5 Hz. Therefore, the wavelet transform scales had the following periods: 0.4 s, 0.8 s, 1.6 s, 3.2 s, 6.4 s, 12.8 s, which corresponded a1, a2, a3, a4, a5, a6 sequences and b1, b2, b3, b4, b5, b6 sequences.

Fig. 3.

Example of Haar wavelet. s, initial signal; a, scaling function; d, detailed oscillations for each of the frequencies.

2.4 Tests for establishing psychological characteristics

The psychological characteristics of the subjects were established using the standard Eysenck test [18], Spielberger test [19], and working memory test. Both Eysenck and Spielberger tests are adapted Russian-language sets of questions, answering which the subjects on specific scales form indicators. The situational and personal anxiety (for the Spielberger test) and indicators of neuroticism and extraversion (for the Eysenck test) are determined. The working memory test is implemented as follows: a 5 × 5 square matrix was set and, in each trial, a selection of three squares was generated, the sequence of which had to be memorized and correctly determined by the user at the end of each trial; the number of trails was ten.

2.5 DNA extraction

Extraction of total DNA from the buccal mucosa was performed using a genome DNA extraction kit (Dia-M, Russia). The quality and purity of DNA were estimated via gel-electrophoresis using a 2% agarose gel in a TAE buffer.

2.6 PCR-RFLP analysis

Genotyping was performed via polymerase chain reaction and restriction fragment length polymorphism (PCR-RFLP) analysis according to previously described methods [9]. PCR was performed using a Bio-Rad CFX96TM system (Bio-Rad, USA) with HS SYBR qPCR mix (Evrogen, Russia). Initial denaturation was carried out at 95 C for 3 min, which was followed by 35 cycles of denaturation at 95 C for 30 s; annealing of primers at 59–71 C for 30 s, elongation at 72 C for 30 s. The PCR product was treated by restriction endonucleases (SibEnzyme, Russia) at 37 C for 2 h, after which deactivation of restriction endonucleases was performed at 65 C for 20 min. The PCR product was then visualized via gel-electrophoresis using a 2% agarose gel in TAE buffer. The primers sequences, conditions of PCR, and the RFLP analysis are presented in Table 1.

Table 1.Condition of restriction fragment length polymorphism analysis.
# Mutation Gene Primers 5′-3′ Annealing t C Restriction endonuclease Product size
TGC↓A A/G–78, 207; 124
TGC↓A C/T–188; 134; 54
GGC↓C C/T–172; 99; 72
GGC↓C A/G–166; 134; 32
CGC↓G T/C–72; 45; 27
AAT↓ATT G/T–89; 212; 301
7 rs4290270 TPH2 F: TTTTGTTTTGGGTGCCATTT 59 C FauND I* T/T–209
GTAT↓AC T/A–209; 134; 76
* 10 mg/mL BSA additionally used.
2.7 Statistical analysis

Statistica 10 software (StatSoft, USA) was used for data analysis. The data were represented as the mean ± standard error mean and Q1; median; Q3. Statistical analysis methods used for descriptive statistics included the Kruskal-Wallace test, Mann-Whitney test, and Friedman tests for paired cases, while the parameter α was taken equal to 5%. To analyze categorical variables, Fisher’s exact test and χ2 test were used. A nonparametric method was used to analyze the relationship between facts—the Spearman’s rank correlation coefficient. The effect of multiple comparisons was taken into account. A multivariate analysis of correspondences was used as an experimental method based on the frequencies of specific observations [20].

3. Results
3.1 Relationship between rs6265 polymorphism and VOI performance

Carriers of the A/A genotype were characterized by a significant standard deviation from the mean in the amplitude of eye movement compared to the G/G genotype during performance of tasks using the VOI, which is best seen on b-sequences (Table 2). At the same time, the mean amplitude of eye movement was also higher (mean values of a-sequences) (Table 3).

Table 2. Values of b-sequences for different scales of discrete wavelet transform (Mean ± S.E.M; Q25; Me; Q75).
Rs6265 A/A A/G G/G
47 82 20
Mean b1 p = 0.0354 –0.01 ± 0.10; –0.17 ± 0.06; –0.35 ± 0.16;
–0.35; 0.09; 0.40 –0.51; –0.18; 0.15 –0.76; –0.24; 0.03
SD b1 p = 0.0022 14.35 ± 0.51; 12.16 ± 0.39; 13.13 ± 1.08;
11.54; 14.32; 16.44 9.88; 11.46; 13.99 8.94; 12.34; 16.81
Mean b2 0.15 ± 0.13; –0.03 ± 0.08; –0.16 ± 0.18;
–0.48; 0.17; 0.67 –0.43; –0.06; 0.49 –0.67; –0.15; 0.15
SD b2 p = 0.0018 12.46 ± 0.48; 10.34 ± 0.39; 11.02 ± 0.90;
10.33; 12.04; 14.97 7.96; 9.61; 12.31 8.06; 10.04; 13.87
Mean b3 0.12 ± 0.16; –0.18 ± 0.12; 0.32 ± 0.22;
–0.47; 0.25; 0.83 –0.71; –0.20; 0.37 –0.35; –0.05; 0.69
SD b3 p = 0.0047 9.99 ± 0.44; 8.21 ± 0.31; 9.02 ± 0.80;
7.63; 9.55; 12.15 6.35; 7.79; 9.94 6.02; 9.95; 11.73
Mean b4 0.01 ± 0.21; 0.00 ± 0.14; –0.02 ± 0.22;
–0.77; –0.34; 1.06 –0.47; 0.02; 0.61 –0.44; 0.05; 0.32
SD b4 p = 0.0000 7.95 ± 0.25; 6.36 ± 0.23; 6.82 ± 0.66;
6.71; 7.89; 9.28 4.85; 6.26; 7.56 4.42; 6.98; 8.99
Mean b5 0.14 ± 0.21; –0.20 ± 0.11; –0.33 ± 0.27;
–0.77; 0.22; 1.15 –0.74; –0.12; 0.53 –1.31; –0.20; 0.27
SD b5 p = 0.0003 6.12 ± 0.26; 4.97 ± 0.21; 4.96 ± 0.50;
5.03; 5.85; 7.20 3.69; 4.52; 5.79 3.50; 4.74; 6.66
Mean b6 –0.18 ± 0.24; 0.17 ± 0.15; 0.24 ± 0.34;
–1.38; –0.10; 1.22 –0.56; 0.04; 0.87 –0.77; –0.12; 1.39
SD b6 p = 0.0013 4.55 ± 0.23; 3.63 ± 0.18; 4.03 ± 0.44;
3.56; 4.23; 5.23 2.44; 3.38; 4.35 2.45; 3.26; 5.29
p-value is indicated only for cases p < 0.05.
Table 3.Values of a–sequences for different scales of discrete wavelet transform (Mean ± S.E.M; Q25; Me; Q75).
Rs6265 A/A A/G G/G
47 82 20
Mean а1 p = 0.0220 11.02 ± 0.79; 8.77 ± 0.36; 9.40 ± 1.16;
7.48; 9.82; 12.83 6.52; 8.31; 9.75 5.62; 7.12; 11.77
SD а1 p = 0.0002 20.62 ± 0.66; 16.95 ± 0.57; 18.11 ± 1.40;
16.94; 19.84; 23.25 13.43; 15.97; 19.63 12.62; 18.62; 22.65
Mean а2 p = 0.0210 10.97 ± 0.80; 8.69 ± 0.36; 9.34 ± 1.17;
7.48; 9.82; 12.66 6.50; 8.32; 9.61 5.42; 7.12; 11.77
SD а2 p = 0.0001 16.21 ± 0.53; 13.12 ± 0.46; 14.08 ± 1.13;
13.41; 15.28; 19.27 10.30; 12.52; 15.23 9.25; 14.65; 17.96
Mean а3 p = 0.0253 10.92 ± 0.80; 8.64 ± 0.35; 9.29 ± 1.17;
7.42; 9.85; 12.66 6.50; 8.29; 9.46 5.22; 7.13; 11.77
SD а3 p = 0.0001 12.54 ± 0.40; 10.12 ± 0.37; 10.49 ± 0.92;
10.64; 12.42; 14.03 7.61; 9.65; 11.74 6.96; 10.40; 13.55
Mean а4 p = 0.0227 10.85 ± 0.80; 8.60 ± 0.36; 9.23 ± 1.19;
7.42; 9.89; 12.66 6.50; 8.10; 9.48 5.13; 7.12; 11.77
SD а4 p = 0.0004 9.54 ± 0.38; 7.77 ± 0.30; 7.88 ± 0.70;
7.72; 9.26; 11.16 5.80; 7.41; 9.01 5.16; 7.13; 10.37
Mean а5 p = 0.0193 10.81 ± 0.81; 8.52 ± 0.35; 9.16 ± 1.18;
7.52; 9.83; 12.79 6.33; 7.99; 9.48 5.13; 6.85; 11.60
SD а5 p = 0.0023 7.21 ± 0.34; 5.88 ± 0.26; 6.04 ± 0.55;
5.60; 6.46; 8.71 4.20; 5.45; 6.87 4.36; 5.25; 8.01
Mean а6 p = 0.0307 10.74 ± 0.80; 8.52 ± 0.36; 9.18 ± 1.21;
7.52; 9.46; 12.79 6.33; 8.04; 9.42 5.13; 6.76; 11.60
SD а6 5.44 ± 0.33; 4.63 ± 0.23; 4.45 ± 0.36;
3.81; 5.15; 7.39 3.24; 4.20; 5.60 3.24; 3.92; 5.47
p value is indicated only for cases p <0.05.

The relationship between the rs6265 polymorphism and goal achievement rates was investigated. It was established (df = 2, p = 0.0017, χ2 criterion = 12.7225) that inferior results (14 experiments with goal achievement, 33 experiments with failure to achieve the goal) were associated with the A/A genotype. The G/G genotype was associated with superior results (15 and 5 experiments for goal achievement and failure, respectively). However, no connection between genotype and the number of errors was found. It is noteworthy that, when the data were divided into two groups (genotypes A/A and A/G versus genotype G/G), there were no differences in the frequency–temporal characteristics of eye movement (p > 0.05), however, users in the first group showed significantly higher values of sincerity (p < 0.05, H (1, N = 149) = 4.542031) (Fig. 4A) and extraversion (p < 0.005, Median Test, Chi-Square = 6.750773 df = 1 p = 0.0094) (Fig. 4B).

When the data were divided into two groups according to another principle (genotype A/A versus genotypes A/G and G/G), differences in eye movement parameters were observed. Differences were significant primarily for increases in the range of eye movement of a–sequences at all studied periods (p < 0.001 for 0.4 s, 0.8 s, 1.6 s, 3.2 s, 6.4 s, 12.8 s) in the first group. For b–sequences, which reflect detailed eye movement features, the range of eye movement was significantly higher in the A/A group for all studied periods, as shown in Fig. 5. The A/G and G/G group achieved the goal significantly more often (H (1, N = 149) = 11.39475 p = 0.0007). Additionally, this group showed inferior results on working memory tests (H (1, N = 149) = 9.265089 p = 0.0023) and lower values of personal anxiety (H (1, N = 149) =8.587107 p = 0.0034).

Fig. 4.

Psychological tests for the different groups of rs6265 genotypes (genotypes A/A and A/G versus genotype G/G). (A) Values of sincerity according to Eysenck test. Kruskal-Wallace test, p < 0.05. (B) Extraversion values on the Eysenck scale. Kruskal-Wallace test, p < 0.005.

Fig. 5.

p values for b–sequences via comparison group 1 (genotype A/A) versus group 2 (genotypes A/G and G/G). Greyline—p- value of differences between the mean of a wavelet transform coefficient. Blackline—p-value of differences between the standard deviation of a wavelet transform coefficient.

3.2 Relationship between rs2030324 polymorphism and VOI performance

Regarding rs2030324, no differences were found when the SNP was divided into three groups (C/C, T/T, and C/T). However, the sincerity values were significantly higher when the combined genotypes C/C and C/T were compared to the T/T genotype (H (1, N = 149) = 9.185538 p = 0.0024) (Supplementary Tables 1 and 2). Extraversion and neuroticism were higher for the T/T genotype (H (1, N = 149) = 9.550368 p = 0.0020 and H (1, N = 149) = 5.586052 p = 0.0181, respectively) on the Eysenck test. A smaller number of goal achievements in the experiment (Median test Chi-Square = 7.143649 df = 1 p = 0.0075) corresponded to higher values of extraversion. However, no dependence was found for rs2030324. Comparing the C/T and T/T genotypes to the C/C genotype did not reveal any significant differences.

3.3 Relationships between other polymorphisms and VOI performance

The rs4570625 polymorphism and eye movement relationship were observed for the 0.4 s, 1.6 s, and 3.2 s periods. The differences between T/G and G/G genotypes were expressed in the change in the standard deviation values (p < 0.01, p < 0.01, p < 0.05, respectively), but not the mean value. There were no differences for more extended periods (Tables 4 and 5). There were no associations with achievement or non-achievement of the goal in the experiment.

Table 4.Values of a-sequences for different scales of discrete wavelet transform (Mean ± S.E.M; Q25; Me; Q75).
Rs4570625 T/G G/G
52 97
Mean a1 11.13 ± 0.87; 8.72 ± 0.27;
6.41; 9.03; 15.90 6.76; 8.44; 10.21
SD a1 p = 0.0081 19.70 ± 0.68; 17.49 ± 0.55;
16.28; 19.21; 22.64 13.43; 16.88; 20.56
Mean a2 11.06 ± 0.88; 8.66 ± 0.27;
6.29; 8.88; 15.91 6.71; 8.39; 10.21
SD a2 p = 0.0036 15.47 ± 0.57; 13.55 ± 0.44;
12.82; 15.16; 17.86 10.35; 12.84; 15.86
Mean a3 10.99 ± 0.88; 8.62 ± 0.27;
6.28; 8.50; 15.91 6.74; 8.41; 10.25
SD a3 p = 0.0395 11.68 ± 0.46; 10.53 ± 0.35;
9.32; 11.55; 13.28 7.53; 10.49; 12.49
Mean a4 10.93 ± 0.87; 8.57 ± 0.27;
6.22; 8.22; 15.91 6.68; 8.41; 10.22
SD a4 8.79 ± 0.36; 8.10 ± 0.30;
6.74; 8.83; 10.29 5.96; 7.76; 9.39
Mean a5 10.88 ± 0.88; 8.49 ± 0.27;
6.18; 8.22; 15.91 6.67; 8.27; 9.95
SD a5 6.75 ± 0.34; 6.09 ± 0.24;
4.81; 6.22; 8.21 4.50; 5.84; 7.10
Mean a6 10.91 ± 0.87; 8.45 ± 0.27;
6.10; 8.22; 15.85 6.68; 8.22; 9.98
SD a6 5.06 ± 0.30; 4.76 ± 0.21;
3.40; 4.41; 6.21 3.35; 4.50; 5.60
p value is indicated only for cases p < 0.05.
Table 5.Values of b-sequences for different scales of discrete wavelet transform (Mean ± S.E.M; Q25; Me; Q75).
Rs4570625 T/G G/G
52 97
Mean b1 –0.26 ± 0.11; –0.08 ± 0.05;
–0.84; –0.22; 0.24 –0.41; –0.01; 0.19
SD b1 p = 0.0188 14.16 ± 0.60; 12.35 ± 0.34;
10.80; 13.28; 17.62 10.07; 12.19; 14.73
Mean b2 –0.00 ± 0.12; 0.02 ± 0.08;
–0.56; 0.00; 0.55 –0.42; –0.04; 0.49
SD b2 11.80 ± 0.47; 10.72 ± 0.37;
9.55; 11.07; 13.87 7.94; 10.33; 12.85
Mean b3 0.04 ± 0.18; –0.05 ± 0.10;
–0.75; –0.06; 0.79 –0.59; –0.02; 0.46
SD b3 p = 0.0005 9.94 ± 0.40; 8.31 ± 0.30;
8.32; 9.97; 11.26 6.20; 7.71; 9.85
Mean b4 –0.05 ± 0.23; 0.04 ± 0.11;
–0.84; –0.14; 0.98 –0.50; –0.08; 0.52
SD b4 p = 0.0207 7.57 ± 0.32; 6.58 ± 0.22;
6.21; 7.12; 8.51 4.88; 6.71; 8.00
Mean b5 –0.41 ± 0.19; 0.05 ± 0.11;
–1.50; –0.42; 0.47 –0.64; 0.02; 0.93
SD b5 5.59 ± 0.24; 5.19 ± 0.21;
4.37; 5.36; 6.50 3.83; 5.03; 6.18
Mean b6 –0.08 ± 0.27; 0.15 ± 0.12;
–1.32; –0.28; 1.11 –0.66; 0.13; 1.09
SD b6 4.37 ± 0.25; 3.77 ± 0.16;
3.20; 4.04; 5.46 2.67; 3.56; 4.41
p -value is indicated only for cases p < 0.05.

rs4290270 did not demonstrate a direct relationship with eye movement during VOI performance. However, this SNP positively and negatively influenced various psychological parameters associated with mastering the interfaces (Fig. 6).

The rs429358 SNP showed a connection exclusively with short-periods (0.4–0.8 s) of eye movements. Dividing the data into two groups for this SNP (C/C genotype versus C/T and T/T genotypes) primarily demonstrated differences in the range of eye movement during VOI performance (p < 0.001) (Fig. 7).

No associations between the rs6313 polymorphism and eye movement parameters were observed during VOI performance. However, the T/T genotype of this SNP was more likely (Chi-square: 6.01494, df = 1, p = 0.014187) not to achieve the experimental task when compared to the combined C/C and C/T genotype group.

Fig. 6.

Results of psychological tests in the different groups of rs4290270 genotypes. (A) Values of working memory. Kruskal-Wallace test, p < 0.001. (B) Extraversion values on the Eysenck scale. Kruskal-Wallace test, p < 0.001.

Fig. 7.

Differences in the range of eye movement during VOI performance in the different groups for the s429358 genotypes. C/C genotype versus C/T and T/T genotypes. (A) The standard deviation of a–sequences for 0.4 s period. Kruskal-Wallace test, p < 0.01. (B) Standard deviation.

3.4 Multivariate analysis of correspondences

Large numbers of differences in various values, including those indirectly affecting mastering the VOI, require multivariate statistical approaches. Therefore, as an exploratory method, we used a multivariate analysis of correspondences, which, based on the frequencies of occurrence of specific observations, finds their coordinates in the space of given dimension, in our case amounting to 13 axes (the total number of SNPs and the result of achieving the goal: “achieved” or “not achieved”) (Fig. 8).

Fig. 8.

A multivariate analysis of correspondences based on the frequencies of occurrence of specific observations finds their coordinates in the space of given dimension, in our case amounting to 13 axes (the total number of SNPs and the result of achieving the goal: “achieved” or “not achieved”).

It is noteworthy that the “achieved” group mainly includes heterozygotes, which are known to have a high adaptive potential due to possessing both allele variants. An exception to this is the rs10119 polymorphism. In contrast, the “not achieved” group generally includes homozygotes, which have a lower adaptive potential than heterozygotes. Again, the rs10119 polymorphism was the exception (Fig. 8).

4. Discussion

rs6265 is the most thoroughly studied SNP in the brain–derived neurotrophic factor (BDNF) gene, released in response to synapse activation and plays a crucial role in maintaining synaptic plasticity [21]. The nucleotide substitution from G to A causes a valine (Val) to methionine (Met) substitution at codon 66 [22]. Carriers of the Met allele have decreased BDNF secretion activity compared to homozygous carriers of the Val allele, which some researchers associate with the low synaptic plasticity of carriers of the Met allele [21].

We have shown that the A/A genotype is associated with a significant standard deviation from the mean in eye movement amplitude. The effects of rs6265 polymorphisms on eye movement patterns have not previously been studied. However, the effects of BDNF on global functional connectivity density in the left frontal eye field have been shown [23].

Previous studies using functional magnetic resonance imaging have shown that the A/A genotype of the BDNF gene is associated with impaired brain motor function, altered short-term plasticity, and more significant errors in short-term motor learning [24]. In addition, the G/A and A/A genotypes of the rs6265 polymorphism increase the risk of posttraumatic stress disorder [25].

We observed that the T/T genotype of rs6313 did not perform the experimental task significantly more often (p < 0.05) than the C/T and C/C genotypes. Earlier results from our group have shown that, on the contrary, carriers of the T allele gave more correct commands. However, these studies involved an electromyographic interface, not VOI [9]. The C allele of the rs6313 polymorphism increases the efficiency of protein translation and concentration of the 5–HT2A receptor, which plays a crucial role in serotonin metabolism [26, 27]. In addition, there is evidence that carriers of the C allele are characterized by increased impulsivity [28, 29]. C allele negatively affects development on the electromyographic interface. However, it helps when performing tasks on the VOI, obviously relating to the different control channels for skeletal muscles and extraocular muscles.

Multivariate analysis of correspondences showed that the “achieved” group mainly includes heterozygotes, which are known to have a high adaptive potential due to possessing both allele variants [30] (Fig. 8). An exception to this is the rs10119 polymorphism, in which the G/G genotype is associated with tremendous success in mastering the VOI. In contrast, the A/A genotype is at risk of developing Alzheimer’s disease [31]. In contrast, the “not achieved” group generally includes homozygotes, which have a lower adaptive potential than heterozygotes (Fig. 8). Again, the rs10119 polymorphism was the exception. However, it should note that this mutation is located in the 3’–untranslated (UTR) region of the gene and affects the efficiency, gene transcription, transcript stability and splicing of the TOMM40 gene [31]. The TOMM40 gene encodes the TOMM40 protein, whose dimers create pores in the outer mitochondrial membrane through which almost all nuclear–encoded proteins enter the mitochondria. Thus, TOMM40 is essential for the biogenesis and functioning of mitochondria, and its impairment is fraught with the development of neurodegenerative diseases [15, 32].

The age of the participants is one of the limitations. The success of VOI mastering is probably because the experiment involved participants aged between 19 to 23 years. However, people with disabilities of different ages. Such studies in older people could broaden our understanding of the genetic determinants of VOI success.

5. Conclusions

Thus, we have shown that the success in the VOI is, at least in part, determined by the users’ genotype. An essential gene is the BDNF gene and the rs6265 polymorphism among the analyzed genes. The product of this gene is significant for working memory and long-term potentiation [33]. The results obtained confirm the assumption that working memory is crucial for developing success in the VOI. These processes are genetically determined to help ensure the optimization of device management training for equipment operators and people with disabilities.


APOE, apolipoprotein E; BDNF, brain-derived neurotrophic factor; DNA, deoxyribonucleic acid; EEG, electroencephalography; HTR2A, 5-hydroxytryptamine receptor 2A; PCR, polymerase chain reaction; PCR-RFLP, polymerase chain reaction-restriction fragment length polymorphism; RFLP analysis, restriction fragment length polymorphism analysis; TAE buffer, Tris-acetate-EDTA buffer; TOMM40, translocase Of Outer Mitochondrial Membrane 40; TPH2, tryptophan hydroxylase 2; SNP, single nucleotide polymorphism; UTR, untranslated; VOI, video-oculographic interface.

Author contributions

These should be presented as follows: YAT, VNP and AAV designed the research. YAT and AAV performed the research using a video-oculographic interface. AAP and IYV performed genetic research. YAT analyzed the data. YAT, AAP, IYV wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate

The Ethics Committee approved the current research of Voronezh State University (Protocol No. 42-03 dated February 13, 2017). All experimental participants gave informed consent for an EEG in accordance with the Helsinki Declaration. Furthermore, informed written consent for enrollment in the research was obtained from all the participants, who were informed of any subsequent genetic testing conducted using their biological material.


We thank four anonymous reviewers for their excellent criticism of the article.


A Russian Foundation for Basic Research funded this research, grant number 17-29-02505 ofi_m to AAV and 19-29-01156 mk to YAT and President grant for support of leading scientific school (Agreement NSh 2535.2020.11) to VNP.

Conflict of interest

Given his role as the Review Board Member of JIN, Dr. Artem P. Gureev had no involvement in the peer-review of this article and has no access to information regarding its peer-review.

Atkinson S, Woods V, Haslam RA, Buckle P. Using non–keyboard input devices: interviews with users in the workplace. International Journal of Industrial Ergonomics. 2004; 33: 571–579.
Turpin G, Armstrong J, Frost P, Fine B, Ward C, Pinnington L. Evaluation of alternative computer input devices used by people with disabilities. Journal of Medical Engineering & Technology. 2005; 29: 119–129.
CS Nam, Nijholt A, Lotte F. Brain—computer interfaces handbook: technological and theoretical advances. New York: CRC Press. 2018.
Igual C, Pardo LA, Hahne J, Igual JM. Myoelectric control for upper limb prostheses. Electronics. 2019; 8: 1244.
Mariano DTG, Freitas AM, Luiz LMD, Silva AN, Pierre P, Naves ELM. An accelerometer–based human computer interface driving an alternative communication system. 5Th ISSNIP–IEEE Biosignals and Biorobotics Conference (2014). 2014; 1–5.
Turovsky YA, Kurgalin SD, Alekseev AV. Analiz dvizheniya glaz cheloveka pri upravlenii samokhodnym shassi s ispolzovaniem sistemy videookulograficheskogo interfeisa [Analysis of human eye movements in the management of self–propelled chassis with video eye–tracking interface system]. Sensornye Sistemy. 2017; 31: 51–58. (In Russian)
Wang X, Bylinskii Z, Castelhano M, Hillis J, Duchowski AT. EMICS’20: eye movements as an interface to cognitive state. Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems. 2020; 1–4.
Rejer I. Genetic algorithms for feature selection for brain—computer interface. International Journal of Pattern Recognition and Artificial Intelligence. 2015; 29: 1559008.
Turovsky YA, Gureev AP, Vitkalova IY, Popov VN. Connection between polymorphisms in HTR2a, TPH2, BDNF, TOMM40 genes and the successful mastering of human—computer interfaces. Journal of Genetics. 2019; 98: 93.
Turovskiy YA, Gureev AP, Vitkalova IY, Chumachenko AG, Popov VN. Correlation of BDNF, tomm40, APOE, AQP5 gene polymorphisms and efficacy of human-computer interfaces. Ulyanovsk Medico-biological Journal. 2020; 4: 110-121. (In Russian)
Turovsky YA, Gureev AP, Vitkalova IY, Popov VN, Vakhtin AA. Genetic features of heart rate variability during working with human–computer interfaces. Experimentalnay Psihologia. 2021.
Bruijniks SJE, van Grootheest G, Cuijpers P, de Kluiver H, Vinkers CH, Peeters F, et al. Working memory moderates the relation between the brain–derived neurotropic factor (BDNF) and psychotherapy outcome for depression. Journal of Psychiatric Research. 2020; 130: 424–432.
Meis S, Endres T, Lessmann V. Postsynaptic BDNF signalling regulates long–term potentiation at thalamo—amygdala afferents. Journal of Physiology. 2012; 590: 193–208.
Schmitt JAJ, Wingen M, Ramaekers JG, Evers EAT, Riedel WJ. Serotonin and human cognitive performance. Current Pharmaceutical Design. 2006; 12: 2473–2486.
Roses A, Sundseth S, Saunders A, Gottschalk W, Burns D, Lutz M. Understanding the genetics of APOE and TOMM40 and role of mitochondrial structure and function in clinical pharmacology of Alzheimer’s disease. Alzheimer’s & Dementia. 2016; 12: 687–694.
Turovsky YA, Kurlagin SD, Alekseev AV. Registration of a computer program GLAZDVIG 2.0. Russia: Patent Russia. No2014661357. November 2014.
Ten Lectures on Wavelets. Daubechies I. A wavelet tour of signal processing. San Diego: Academic Press. 1992.
Gomez R, Cooper A, Gomez A. Susceptibility to positive and negative mood states: test of Eysenck’s, Gray’s and Newman’s theories. Personality and Individual Differences. 2000; 29: 351–365.
Spielberger CD. Test Anxiety Inventory. In Weiner, IB and Craighead, WE (eds.) The corsini encyclopedia of psychology (pp. 1–1). Hoboken: John Wiley & Sons. 2010.
Clausen SE. Applied Correspondence Analysis: an introduction. California: SAGE Publishers. 1998.
Baj G, Carlino D, Gardossi L, Tongiorgi E. Toward a unified biological hypothesis for the BDNF Val66Met–associated memory deficits in humans: a model of impaired dendritic mRNA trafficking. Frontiers in Neuroscience. 2013; 7: 188.
Sheikh HI, Hayden EP, Kryski KR, Smith HJ, Singh SM. Genotyping the BDNF rs6265 (val66met) polymorphism by one–step amplified refractory mutation system PCR. Psychiatric Genetics. 2010; 20: 109–112.
Li W, Liu B, Xu J, Jiang T, Yu C. Interaction of COMT rs4680 and BDNF rs6265 polymorphisms on functional connectivity density of the left frontal eye field in healthy young adults. Human Brain Mapping. 2016; 37: 2468–2478.
McHughen SA, Rodriguez PF, Kleim JA, Kleim ED, Crespo LM, Procaccio V, et al. BDNF val66met polymorphism influences motor system function in the human brain. Cerebral Cortex. 2009; 20: 1254–1262.
Guo J, Yang Y, Zheng J, Guo M, Wang X, Gao Y, et al. Functional rs6265 polymorphism in the brain‐derived neurotrophic factor gene confers protection against neurocognitive dysfunction in posttraumatic stress disorder among Chinese patients with hepatocellular carcinoma. Journal of Cellular Biochemistry. 2019; 20: 10434–10443.
Smith RM, Papp AC, Webb A, Ruble CL, Munsie LM, Nisenbaum LK, et al. Multiple regulatory variants modulate expression of 5–hydroxytryptamine 2a receptors in human cortex. Biological Psychiatry. 2013; 73: 546–554.
Smith RM, Banks W, Hansen E, Sadee W, Herman GE. Family–based clinical associations and functional characterization of the serotonin 2a receptor gene (HTR2A) in autism spectrum disorder. Autism Research. 2014; 7: 459–467.
Racine SE, Culbert KM, Larson CL, Klump KL. The possible influence of impulsivity and dietary restraint on associations between serotonin genes and binge eating. Journal of Psychiatric Research. 2009; 43: 1278–1286.
Jakubczyk A, Wrzosek M, Lukaszkiewicz J, Sadowska–Mazuryk J, Matsumoto H, Sliwerska E, et al. The CC genotype in HTR2A T102C polymorphism is associated with behavioral impulsivity in alcohol–dependent patients. Journal of Psychiatric Research. 2012; 46: 44–49.
Sellis D, Kvitek DJ, Dunn B, Sherlock G, Petrov DA. Heterozygote advantage is a common outcome of adaptation in saccharomyces cerevisiae. Genetics. 2016; 203: 1401–1413.
Zhu Z, Yang Y, Xiao Z, Zhao Q, Wu W, Liang X, et al. TOMM40 and APOE variants synergistically increase the risk of Alzheimer’s disease in a Chinese population. Aging Clinical and Experimental Research. 2020.
Bender A, Desplats P, Spencer B, Rockenstein E, Adame A, Elstner M, et al. TOM40 mediates mitochondrial dysfunction induced by α–synuclein accumulation in Parkinson’s disease. PLoS ONE. 2013; 8: e62277.
Gureev AP, Popov VN, Starkov AA. Crosstalk between the mTOR and Nrf2/ARE signaling pathways as a target in the improvement of long–term potentiation. Experimental Neurology. 2020; 328: 113285.
Back to top