Reducing the Time Requirement of k-Means Algorithm

Traditional k-means and most k-means variants are still computationally expensive for large datasets, such as microarray data, which have large datasets with large dimension size d. In k-means clustering, we are given a set of n data points in d-dimensional space R^d and an integer k. The problem is to determine a set of k points in R^d, called centers, so as to minimize the mean squared distance from each data point to its nearest center. In this work, we develop a novel k-means algorithm, which is simple but more efficient than the traditional k-means and the recent enhanced k-means. Our new algorithm is based on the recently established relationship between principal component analysis and the k-means clustering. We provided the correctness proof for this algorithm. Results obtained from testing the algorithm on three biological data and six non-biological data (three of these data are real, while the other three are simulated) also indicate that our algorithm is empirically faster than other known k-means algorithms. We assessed the quality of our algorithm clusters against the clusters of a known structure using the Hubert-Arabie Adjusted Rand index (ARI_HA). We found that when k is close to d, the quality is good (ARI_HA>0.8) and when k is not close to d, the quality of our new k-means algorithm is excellent (ARI_HA>0.9). In this paper, emphases are on the reduction of the time requirement of the k-means algorithm and its application to microarray data due to the desire to create a tool for clustering and malaria research. However, the new clustering algorithm can be used for other clustering needs as long as an appropriate measure of distance between the centroids and the members is used. This has been demonstrated in this work on six non-biological data.     

Derivative component analysis for mass spectral serum proteomic profiles

Background

As a promising way to transform medicine, mass spectrometry based proteomics technologies have seen a great progress in identifying disease biomarkers for clinical diagnosis and prognosis. However, there is a lack of effective feature selection methods that are able to capture essential data behaviors to achieve clinical level disease diagnosis. Moreover, it faces a challenge from data reproducibility, which means that no two independent studies have been found to produce same proteomic patterns. Such reproducibility issue causes the identified biomarker patterns to lose repeatability and prevents it from real clinical usage.

Methods

In this work, we propose a novel machine-learning algorithm: derivative component analysis (DCA) for high-dimensional mass spectral proteomic profiles. As an implicit feature selection algorithm, derivative component analysis examines input proteomics data in a multi-resolution approach by seeking its derivatives to capture latent data characteristics and conduct de-noising. We further demonstrate DCA's advantages in disease diagnosis by viewing input proteomics data as a profile biomarker via integrating it with support vector machines to tackle the reproducibility issue, besides comparing it with state-of-the-art peers.

Results

Our results show that high-dimensional proteomics data are actually linearly separable under proposed derivative component analysis (DCA). As a novel multi-resolution feature selection algorithm, DCA not only overcomes the weakness of the traditional methods in subtle data behavior discovery, but also suggests an effective resolution to overcoming proteomics data's reproducibility problem and provides new techniques and insights in translational bioinformatics and machine learning. The DCA-based profile biomarker diagnosis makes clinical level diagnostic performances reproducible across different proteomic data, which is more robust and systematic than the existing biomarker discovery based diagnosis.

Conclusions

Our findings demonstrate the feasibility and power of the proposed DCA-based profile biomarker diagnosis in achieving high sensitivity and conquering the data reproducibility issue in serum proteomics. Furthermore, our proposed derivative component analysis suggests the subtle data characteristics gleaning and de-noising are essential in separating true signals from red herrings for high-dimensional proteomic profiles, which can be more important than the conventional feature selection or dimension reduction. In particular, our profile biomarker diagnosis can be generalized to other omics data for derivative component analysis (DCA)'s nature of generic data analysis.

     

Defining and Evaluating Classification Algorithm for High-Dimensional Data Based on Latent Topics

Automatic text categorization is one of the key techniques in information retrieval and the data mining field. The classification is usually time-consuming when the training dataset is large and high-dimensional. Many methods have been proposed to solve this problem, but few can achieve satisfactory efficiency. In this paper, we present a method which combines the Latent Dirichlet Allocation (LDA) algorithm and the Support Vector Machine (SVM). LDA is first used to generate reduced dimensional representation of topics as feature in VSM. It is able to reduce features dramatically but keeps the necessary semantic information. The Support Vector Machine (SVM) is then employed to classify the data based on the generated features. We evaluate the algorithm on 20 Newsgroups and Reuters-21578 datasets, respectively. The experimental results show that the classification based on our proposed LDA+SVM model achieves high performance in terms of precision, recall and F1 measure. Further, it can achieve this within a much shorter time-frame. Our process improves greatly upon the previous work in this field and displays strong potential to achieve a streamlined classification process for a wide range of applications.     

Efficient Deterministic Finite Automata Minimization Based on Backward Depth Information

Obtaining a minimal automaton is a fundamental issue in the theory and practical implementation of deterministic finite automatons (DFAs). A minimization algorithm is presented in this paper that consists of two main phases. In the first phase, the backward depth information is built, and the state set of the DFA is partitioned into many blocks. In the second phase, the state set is refined using a hash table. The minimization algorithm has a lower time complexity O(n) than a naive comparison of transitions O(n²). Few states need to be refined by the hash table, because most states have been partitioned by the backward depth information in the coarse partition. This method achieves greater generality than previous methods because building the backward depth information is independent of the topological complexity of the DFA. The proposed algorithm can be applied not only to the minimization of acyclic automata or simple cyclic automata, but also to automata with high topological complexity. Overall, the proposal has three advantages: lower time complexity, greater generality, and scalability. A comparison to Hopcroft’s algorithm demonstrates experimentally that the algorithm runs faster than traditional algorithms.     

The Effectiveness of Voriconazole in Therapy of <Emphasis Type="Italic">Candida glabrata’s</Emphasis> Biofilms Oral Infections and Its Influence on the Matrix Composition and Gene Expression

Candida glabrata is one of most prevalent yeast in fungal infections, especially in immunocompromised patients. Its azole resistance results in a low therapeutic response, particularly when associated with biofilms. The main goal of this work was to study the effectiveness of voriconazole (Vcz) against C. glabrata biofilms oral pathologies, as esophageal or oropharyngeal candidiasis. Antifungal susceptibilities were determined in pre-formed 24-h-biofilms and ERG genes expression was determined by qRT-PCR. Protein quantification was performed using BCA^® Kit, carbohydrate was estimated according to the Dubois assay and β-1,3 glucans concentration were determined using Glucatell^® kit. Finally, ergosterol, Vcz, and fluconazole (Flu) concentrations within the biofilm matrices were determined by RP-HPLC. Results showed that C. glabrata biofilms were more susceptible to Vcz than to Flu and that ERG genes expression evidenced an overexpression of the three ERG genes in the presence of both azoles. The matrix content presented a remarked decrease in proteins and an increase in carbohydrates, namely β-1,3 glucans. Ergosterol was successfully detected and quantified in the biofilm matrices, with no differences in all the considered conditions. Vcz demonstrated better diffusion through the biofilms and better cell penetration capacities, than Flu, indicating that the structure of the drug molecule fully influences its dissemination through the biofilm matrices. This work showed that Vcz is notably more effective than Flu for the treatment of resistant C. glabrata oral biofilms, which demonstrates a clinical relevance in its future use for the treatment of oropharyngeal/esophageal candidiasis caused by this species.     

拽线法:一个构建系统发育树的新算法

这篇文章要讨论的拽线法（DL）是贪婪算法的一种。和Fitch—Margoliash（FM）一样,DL也是基于距离矩阵构建系统发育树,但是和FM算法相比,DL具有低复杂度、较高的容错性和准确度高的优点。当存在误差时,DL算法只是加大了不在同一个父节点下的基因序列的距离,但能够准确的判断序列的亲缘关系,进而得到完美的进化树拓扑结构;相比之下,FM算法让各个基因序列间的距离均摊了这种误差,从而有可能将本应该具有相同父节点的基因序列分到不同的分支。     

Integration of the M6 Cyberknife in the Moderato Monte Carlo platform and prediction of beam parameters using machine learning

PurposeThis work describes the integration of the M6 Cyberknife in the Moderato Monte Carlo platform, and introduces a machine learning method to accelerate the modelling of a linac.MethodsThe MLC-equipped M6 Cyberknife was modelled and integrated in Moderato, our in-house platform offering independent verification of radiotherapy dose distributions. The model was validated by comparing TPS dose distributions with Moderato and by film measurements. Using this model, a machine learning algorithm was trained to find electron beam parameters for other M6 devices, by simulating dose curves with varying spot size and energy. The algorithm was optimized using cross-validation and tested with measurements from other institutions equipped with a M6 Cyberknife.ResultsOptimal agreement in the Monte Carlo model was reached for a monoenergetic electron beam of 6.75 MeV with Gaussian spatial distribution of 2.4 mm FWHM. Clinical plan dose distributions from Moderato agreed within 2% with the TPS, and film measurements confirmed the accuracy of the model. Cross-validation of the prediction algorithm produced mean absolute errors of 0.1 MeV and 0.3 mm for beam energy and spot size respectively. Prediction-based simulated dose curves for other centres agreed within 3% with measurements, except for one device where differences up to 6% were detected.ConclusionsThe M6 Cyberknife was integrated in Moderato and validated through dose re-calculations and film measurements. The prediction algorithm was successfully applied to obtain electron beam parameters for other M6 devices. This method would prove useful to speed up modelling of new machines in Monte Carlo systems.     

A neural network study on the dynamic identification of a fermentation system

The objective of this paper is to propose neural networks for the study of dynamic identification and prediction of a fermentation system which produces mainly 2,3-butanediol (2,3-BDL). The metabolic products of the fermentation, acetic acid, acetoin, ethanol, and 2,3-BDL were measured on-line via a mass spectrometer modified by the insertion of a dimethylvinylsilicone membrane probe. The measured data at different sampling times were included as the input and output nodes, at different learning batches, of the network. A fermentation system is usually nonlinear and dynamic in nature. Measured fermentation data obtained from the complex metabolic pathways are often difficult to be entirely included in a static process model, therefore, a dynamic model was suggested instead. In this work, neural networks were provided by a dynamic learning and prediction process that moved along the time sequence batchwise. In other words, a scheme of two-dimensional moving window (number of input nodes by the number of training data) was proposed for reading in new data while forgetting part of the old data. Proper size of the network including proper number of input/output nodes were determined by trained with the real-time fermentation data. Different number of hidden nodes under the consideration of both learning performance and computation efficiency were tested. The data size for each learning batch was determined. The performance of the learning factors such as the learning coefficient η and the momentum term coefficient α were also discussed. The effect of different dynamic learning intervals, with different starting points and the same ending point, both on the learning and prediction performance were studied. On the other hand, the effect of different dynamic learning intervals, with the same starting point and different ending points, was also investigated. The size of data sampling interval was also discussed. The performance from four different types of transfer functions, x/(1+|x|), sgn(x)·x ²/(1+x ²), 2/(1+e ^{? x} )?1, and 1/(1+e ^{? x} ) was compared. A scaling factor b was added to the transfer function and the effect of this factor on the learning was also evaluated. The prediction results from the time-delayed neural networks were also studied.     

SVM2Motif—Reconstructing Overlapping DNA Sequence Motifs by Mimicking an SVM Predictor

Identifying discriminative motifs underlying the functionality and evolution of organisms is a major challenge in computational biology. Machine learning approaches such as support vector machines (SVMs) achieve state-of-the-art performances in genomic discrimination tasks, but—due to its black-box character—motifs underlying its decision function are largely unknown. As a remedy, positional oligomer importance matrices (POIMs) allow us to visualize the significance of position-specific subsequences. Although being a major step towards the explanation of trained SVM models, they suffer from the fact that their size grows exponentially in the length of the motif, which renders their manual inspection feasible only for comparably small motif sizes, typically k ≤ 5. In this work, we extend the work on positional oligomer importance matrices, by presenting a new machine-learning methodology, entitled motifPOIM, to extract the truly relevant motifs—regardless of their length and complexity—underlying the predictions of a trained SVM model. Our framework thereby considers the motifs as free parameters in a probabilistic model, a task which can be phrased as a non-convex optimization problem. The exponential dependence of the POIM size on the oligomer length poses a major numerical challenge, which we address by an efficient optimization framework that allows us to find possibly overlapping motifs consisting of up to hundreds of nucleotides. We demonstrate the efficacy of our approach on a synthetic data set as well as a real-world human splice site data set.     

Geochemical Evidence of the Seasonality,Affinity and Pigmenation of Solenopora jurassica

Solenopora jurassica is a fossil calcareous alga that functioned as an important reef-building organism during the Palaeozoic. It is of significant palaeobiological interest due to its distinctive but poorly understood pink and white banding. Though widely accepted as an alga there is still debate over its taxonomic affinity, with recent work arguing that it should be reclassified as a chaetetid sponge. The banding is thought to be seasonal, but there is no conclusive evidence for this. Other recent work has, however demonstrated the presence of a unique organic boron-containing pink/red pigment in the pink bands of S. jurassica. We present new geochemical evidence concerning the seasonality and pigmentation of S. jurassica. Seasonal growth cycles are demonstrated by X-ray radiography, which shows differences in calcite density, and by varying δ¹³C composition of the bands. Temperature variation in the bands is difficult to constrain accurately due to conflicting patterns arising from Mg/Ca molar ratios and δ¹⁸O data. Fluctuating chlorine levels indicate increased salinity in the white bands, when combined with the isotope data this suggests more suggestive of marine conditions during formation of the white band and a greater freshwater component (lower chlorinity) during pink band precipitation (δ¹⁸O). Increased photosynthesis is inferred within the pink bands in comparison to the white, based on δ¹³C. Pyrolysis Gas Chromatography Mass Spectrometry (Py-GCMS) and Fourier Transform Infrared Spectroscopy (FTIR) show the presence of tetramethyl pyrrole, protein moieties and carboxylic acid groups, suggestive of the presence of the red algal pigment phycoerythrin. This is consistent with the pink colour of S. jurassica. As phycoerythrin is only known to occur in algae and cyanobacteria, and no biomarker evidence of bacteria or sponges was detected we conclude S. jurassica is most likely an alga. Pigment analysis may be a reliable classification method for fossil algae.     

Sur l'espace topologique lié à une nouvelle théorie de l'apprentissage

There have been two contrasting doctrines concerning learning, more generally about acquisition of knowledge: empiricism and rationalism. The theory of learning in such a field as artificial intelligence seems to fall within the empiricist framework. On the hand, N. Chomsky and his followers have discussed, during the last decade, concerning learning, especially about language learning, from the rationalist point of view (Chomsky, 1965). The main feature in the rationalist approach toward a theory of learning lies in the speculation that in order to acquire knowledge it is indispensable for a learner to be endowed with “innate ideas”, and that “experience” in the external world are merely subsidiary types of information for the learner. If this is acceptable, we can inquire: Under what kind of innate ideas can the learner understand the structure of the external world? In our previous paper (Uesaka, Aizawa, Ebara, and Ozeki, 1973), we formalized this by introducing the mathematical notion of “learnability”, and gave a partial answer to the above inquiry. In this formalization we assumed that the set F of objects to be learned consists of mappings of N to itself, where N is the set of positive integers. Then, constructing a topological space (F, \(\mathcal{O}\) ) by an appropriate family \(\mathcal{O}\) of open sets, we observed that the notion of learnability can be well described in terms of topological properties of the learning space (F, \(\mathcal{O}\) ). Many problems must be solved, however, before we raise the theory to a complete model of the rationalist theory of learning. The topological study of the space (F, \(\mathcal{O}\) ) is, we believe, the first step toward this approach. In this context, we discuss the topological aspects of this space. Now we define \(\mathcal{O}\) as follows: By N ² we mean the direct product of two N's. Let s be a subset of N ². If, for any (x, y), (x′, y′) in s, x=x′ implies y=y′, then we say that s is single-valued. Let f ∈ F, If, for any (x, y) in s, y=f(x), then f is said to be on s, denoted as \(f\underline \supseteq s\) . Let \(\pi \left( s \right) = \left\{ {g;g \in F,g\underline \supseteq s} \right\}\) . A single-valued finite subset of N ² is called datum. Let D denote the family of all data. Let \(\mathcal{O}* = \left\{ \phi \right\} \cup \left\{ {\pi \left( d \right);d \in D} \right\}\) , and \(\mathcal{O}\) denote the family of all subsets of F, each of which is written as \(\mathop \cup \limits_\alpha W_{\alpha }\) , where W _α is in \(\mathcal{O}*\) . Then, it is easily seen that \(\mathcal{O}\) satisfies the axiom of the open system of a topological space. It is shown that the learning space (F, \(\mathcal{O}\) ) has the following properties:

1. It satisfies the first and the second countability axioms.
2. It is separable and is totally disconnected.
3. It is a Hausdorff space and, further, is regular and normal.
4. It is neither compact nor locally compact.
5. It is metrizable, or more precisely there exists a complete but not totally bounded metric space which is homeomorphic to learning space.
6. Any of its subspace can be embedded into its special subspace.

     

A general method for fast multiple sequence alignment

We have developed a fast heuristic algorithm for multiple sequence alignment which provides near-to-optimal results for sufficiently homologous sequences. The algorithm makes use of the standard dynamic programming procedure by applying it to all pairs of sequences. The resulting score matrices for pair-wise alignment give rise to secondary matrices containing the additional charges imposed by forcing the alignment path to run through a particular vertex. Such a constraint corresponds to slicing the sequences at the positions defining that vertex, and aligning the remaining pairs of prefix and suffix sequences separately. From these secondary matrices, one can compute - for any given family of sequences - suitable positions for cutting all of these sequences simultaneously, thus reducing the problem of aligning a family of n sequences of average length l in a Divide and Conquer fashion to aligning two families of n sequences of approximately half that length.In this paper, we explain the method for the case of 3 sequences in detail, and we demonstrate its potential and its limits by discussing its behaviour for several test families. A generalization for aligning more than 3 sequences is lined out, and some actual alignments constructed by our algorithm for various user-defined parameters are presented.     

Estimating and comparing thermal performance curves

I show how one can estimate the shape of a thermal performance curve using information theory. This approach ranks plausible models by their Akaike information criterion (AIC), which is a measure of a model's ability to describe the data discounted by the model's complexity. I analyze previously published data to demonstrate how one applies this approach to describe a thermal performance curve. This exemplary analysis produced two interesting results. First, a model with a very high r² (a modified Gaussian function) appeared to overfit the data. Second, the model favored by information theory (a Gaussian function) has been used widely in optimality studies of thermal performance curves. Finally, I discuss the choice between regression and ANOVA when comparing thermal performance curves and highlight a superior method called template mode of variation. Much progress can be made by abandoning traditional methods for a method that combines information theory with template mode of variation.     

Nestedness and τ-temperature in ecological networks

The τ-temperature is a measure of disorder of bipartite networks that is based on the total Manhattan distance of the adjacency matrix. Two properties of this measure are that it does not depend on permutations of lines or columns that have the same connectivity and it is completely determined by connectivities of lines and columns. The normalisation of τ is done by an uniform random matrix whose elements were previously sorted. τ shows no bias against uniform random matrices of several occupations, ρ, sizes, L, and shapes. The scaling of the total Manhattan distance of a random matrix is D_rand ∝ L³ρ while the same scaling for a full nested matrix is D_nest ∝ L³ρ^3/2. We test τ for a large set of empirical matrices to verify these scalings. The index τ correlates better with the temperature of Atmar than with the NODF index of nestedness. We conclude this work by discussing differences between nestedness indices and order/disorder indices.     

Multivariate Bayesian analysis of Gaussian,right censored Gaussian,ordered categorical and binary traits using Gibbs sampling

A fully Bayesian analysis using Gibbs sampling and data augmentation in a multivariate model of Gaussian, right censored, and grouped Gaussian traits is described. The grouped Gaussian traits are either ordered categorical traits (with more than two categories) or binary traits, where the grouping is determined via thresholds on the underlying Gaussian scale, the liability scale. Allowances are made for unequal models, unknown covariance matrices and missing data. Having outlined the theory, strategies for implementation are reviewed. These include joint sampling of location parameters; efficient sampling from the fully conditional posterior distribution of augmented data, a multivariate truncated normal distribution; and sampling from the conditional inverse Wishart distribution, the fully conditional posterior distribution of the residual covariance matrix. Finally, a simulated dataset was analysed to illustrate the methodology. This paper concentrates on a model where residuals associated with liabilities of the binary traits are assumed to be independent. A Bayesian analysis using Gibbs sampling is outlined for the model where this assumption is relaxed.     

Adaptive compressive learning for prediction of protein-protein interactions from primary sequence

Protein-protein interactions (PPIs) play an important role in biological processes. Although much effort has been devoted to the identification of novel PPIs by integrating experimental biological knowledge, there are still many difficulties because of lacking enough protein structural and functional information. It is highly desired to develop methods based only on amino acid sequences for predicting PPIs. However, sequence-based predictors are often struggling with the high-dimensionality causing over-fitting and high computational complexity problems, as well as the redundancy of sequential feature vectors. In this paper, a novel computational approach based on compressed sensing theory is proposed to predict yeast Saccharomyces cerevisiae PPIs from primary sequence and has achieved promising results. The key advantage of the proposed compressed sensing algorithm is that it can compress the original high-dimensional protein sequential feature vector into a much lower but more condensed space taking the sparsity property of the original signal into account. What makes compressed sensing much more attractive in protein sequence analysis is its compressed signal can be reconstructed from far fewer measurements than what is usually considered necessary in traditional Nyquist sampling theory. Experimental results demonstrate that proposed compressed sensing method is powerful for analyzing noisy biological data and reducing redundancy in feature vectors. The proposed method represents a new strategy of dealing with high-dimensional protein discrete model and has great potentiality to be extended to deal with many other complicated biological systems.     

Interaction between the Microbial Community and Invading Escherichia coli O157:H7 in Soils from Vegetable Fields

The survival of Escherichia coli O157:H7 in soils can contaminate vegetables, fruits, drinking water, etc. However, data on the impact of E. coli O157:H7 on soil microbial communities are limited. In this study, we monitored the changes in the indigenous microbial community by using the phospholipid fatty acid (PLFA) method to investigate the interaction of the soil microbial community with E. coli O157:H7 in soils. Simple correlation analysis showed that the survival of E. coli O157:H7 in the test soils was negatively correlated with the ratio of Gram-negative (G⁻) to Gram-positive (G⁺) bacterial PLFAs (G⁻/G⁺ ratio). In particular, levels of 14 PLFAs were negatively correlated with the survival time of E. coli O157:H7. The contents of actinomycetous and fungal PLFAs in the test soils declined significantly (P, <0.05) after 25 days of incubation with E. coli O157:H7. The G⁻/G⁺ ratio declined slightly, while the ratio of bacterial to fungal PLFAs (B/F ratio) and the ratio of normal saturated PLFAs to monounsaturated PLFAs (S/M ratio) increased, after E. coli O157:H7 inoculation. Principal component analysis results further indicated that invasion by E. coli O157:H7 had some effects on the soil microbial community. Our data revealed that the toxicity of E. coli O157:H7 presents not only in its pathogenicity but also in its effect on soil microecology. Hence, close attention should be paid to the survival of E. coli O157:H7 and its potential for contaminating soils.     

Measuring Symmetry,Asymmetry and Randomness in Neural Network Connectivity

Cognitive functions are stored in the connectome, the wiring diagram of the brain, which exhibits non-random features, so-called motifs. In this work, we focus on bidirectional, symmetric motifs, i.e. two neurons that project to each other via connections of equal strength, and unidirectional, non-symmetric motifs, i.e. within a pair of neurons only one neuron projects to the other. We hypothesise that such motifs have been shaped via activity dependent synaptic plasticity processes. As a consequence, learning moves the distribution of the synaptic connections away from randomness. Our aim is to provide a global, macroscopic, single parameter characterisation of the statistical occurrence of bidirectional and unidirectional motifs. To this end we define a symmetry measure that does not require any a priori thresholding of the weights or knowledge of their maximal value. We calculate its mean and variance for random uniform or Gaussian distributions, which allows us to introduce a confidence measure of how significantly symmetric or asymmetric a specific configuration is, i.e. how likely it is that the configuration is the result of chance. We demonstrate the discriminatory power of our symmetry measure by inspecting the eigenvalues of different types of connectivity matrices. We show that a Gaussian weight distribution biases the connectivity motifs to more symmetric configurations than a uniform distribution and that introducing a random synaptic pruning, mimicking developmental regulation in synaptogenesis, biases the connectivity motifs to more asymmetric configurations, regardless of the distribution. We expect that our work will benefit the computational modelling community, by providing a systematic way to characterise symmetry and asymmetry in network structures. Further, our symmetry measure will be of use to electrophysiologists that investigate symmetry of network connectivity.     

To what extent naringenin binding and membrane depolarization shape mitoBK channel gating—A machine learning approach

The large conductance voltage- and Ca²⁺-activated K⁺ channels from the inner mitochondrial membrane (mitoBK) are modulated by a number of factors. Among them flavanones, including naringenin (Nar), arise as a promising group of mitoBK channel regulators from a pharmacological point of view. It is well known that in the presence of Nar the open state probability (p_op) of mitoBK channels significantly increases. Nevertheless, the molecular mechanism of the mitoBK-Nar interactions remains still unrevealed. It is also not known whether the effects of naringenin administration on conformational dynamics can resemble those which are exerted by the other channel-activating stimuli. In aim to answer this question, we examine whether the dwell-time series of mitoBK channels which were obtained at different voltages and Nar concentrations (yet allowing to reach comparable p_ops) are discernible by means of artificial intelligence methods, including k-NN and shapelet learning. The obtained results suggest that the structural complexity of the gating dynamics is shaped both by the interaction of channel gate with the voltage sensor (VSD) and the Nar-binding site. For a majority of data one can observe stimulus-specific patterns of channel gating. Shapelet algorithm allows to obtain better prediction accuracy in most cases. Probably, because it takes into account the complexity of local features of a given signal. About 30% of the analyzed time series do not sufficiently differ to unambiguously distinguish them from each other, which can be interpreted in terms of the existence of the common features of mitoBK channel gating regardless of the type of activating stimulus. There exist long-range mutual interactions between VSD and the Nar-coordination site that are responsible for higher levels of Nar-activation (Δp_op) at deeply depolarized membranes. These intra-sensor interactions are anticipated to have an allosteric nature.