Feature selection has been extensively applied to identify cancer genes using omics data. Although substantial studies have been conducted to search for cancer genes, the available rich knowledge on various cancers is seldom used as prior information in feature selection. This paper proposes a two-stage prior LASSO (TSPLASSO) method, which represents an early attempt in designing feature selection algorithms using prior information. The first stage performs gene selection via linear regression with LASSO. Candidate genes that are correlated with known cancer genes are retained for subsequent analysis. The second stage establishes a logistic regression model with LASSO to realize final cancer gene selection and sample classification. The key advantages of TSPLASSO include the successive consideration of prior cancer genes and binary sample types as response variables in stages one and two, respectively. In addition, the TSPLASSO performs sample classification and variable selection simultaneously. Compared with six state-of-the-art algorithms, numerical simulations in six real-world datasets show that TSPLASSO can improve the accuracy of variable selection by 5%-400% in the three bulk sequencing datasets and the scRNA-seq dataset; and the performance is robust against data noise and variations of prior cancer genes. The TSPLASSO provides an efficient, stable and practical algorithm for exploring biomedcial and health informatics from omics data.

MOTIVATION: Both β-value and M-value have been used as metrics to measure methylation levels. The M-value is more statistically valid for the differential analysis of methylation levels. However, the β-value is much more biologically interpretable and needs to be reported when M-value method is used for conducting differential methylation analysis. There is an urgent need to know how to interpret the degree of differential methylation from the M-value. In M-value linear regression model, differential methylation M-value ΔM can be easily obtained from the coefficient estimate, but it is not straightforward to get the differential methylation β-value, Δβ since it cannot be obtained from the coefficient alone. RESULTS: To fill the gap, we have built a bridge to connect the statistically sound M-value linear regression model and the biologically interpretable Δβ. In this article, three methods were proposed to calculate differential methylation values, Δβ from M-value linear regression model and compared with the Δβ directly obtained from β-value linear regression model. We showed that under the condition that M-value linear regression model is correct, the method M-model-coef is the best among the four methods. M-model-M-mean method works very well too. If the coefficients α0, α2,…αp are not given (as 'MethLAB' package), the M-model-M-mean method should be used. The Δβ directly obtained from β-value linear regression model can give very biased results, especially when M-values are not in (-2, 2) or β-values are not in (0.2, 0.8). AVAILABILITY AND IMPLEMENTATION: The dataset for example is available at the National Center for Biotechnology Information Gene Expression Omnibus repository, GSE104778. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Summary Fingerprinting positioning is used for indoor location estimation because of low cost and open access properties, in which a radio map is built by calibrating signal‐strength values at several training locations in the offline phase. However, due to the sophisticated propagation of radio signals, the received signal strength (RSS) in wireless network change as the environment changes, and the radio map built in the offline phase may be out of date. Furthermore, the recalibration of signal‐strength values for each environment change is laborious and time consuming. In this paper, we present a novel algorithm to reconstruct a radio map using real‐time signal‐strength readings received at some reference points. We first demonstrate that different features of signal propagation are obtained in different regions of indoor environment. Then, the indoor environment is divided into several regions by clustering the path‐loss parameters of each reference point. In addition, the relationship between the RSS of reference points and calibration nodes is established with robust linear regression. Finally, the real‐time radio map is updated dynamically according to robust regression and the real‐time RSSs of the calibration nodes. The experimental results show the usefulness of the proposed method and the accuracy of the localization can be improved.

Structural controllability of complex networks has been a research focus in recent years. However, few works considered the structural controllability of biological networks, especially for dynamic biological networks. In this paper, structural controllability of one static and five dynamic transcriptional regulatory networks (TRNs) for Saccharomyces cerevisiae (S. cerevisiae) have been investigated. The five dynamic networks included two endogenous and three exogenous ones. We clarified the controllability properties of these networks, as well as explored the differences among different types of networks. Our results revealed that the structural properties of different types of networks are significantly different. We found that the TRNs are rather difficult to be fully controlled. However, the dynamic TRNs were relatively easier to be controlled than the static one, and one needs relatively more external inputs to control the exogenous TRNs than the endogenous ones. Based on the structural controllability analysis of subnetworks with the same sizes, as well as randomly perturbed networks by preserving degree distributions, we further illustrated that the structural controllability of the six networks may be mainly dominated by their degree distributions. Our investigations clarify the control properties of TRNs in S. cerevisiae, it provides some insights on real-world control of biological networks.

Identification of important nodes is an emerging hot topic in complex networks over the last few decades. The so-called important nodes are hub, influential nodes, leaders, and so on. To characterize the importance of nodes, various indexes are introduced in complex networks, such as degree, closeness, betweenness, k-shell, and principal component analysis based on the adjacency matrix. By using the above indexes and multivariate statistical analysis technique, this paper aims at developing a new approach to identify the important nodes in artificial bio-molecular networks generated from the duplication-divergence (DD) model. In particular, the statistical characteristics of important nodes are also investigated. The above results shed light on the potential real-world applications in bio-molecular networks, such as deducing the genes related to the specific disease.

The performance of Wiener filters in restoring the quality and intelligibility of noisy speech depends on: (i) the accuracy of the estimates of the power spectra or the correlation values of the noise and the speech processes, and (ii) on the Wiener filter structure. In this paper a Bayesian method is proposed where model combination and model decomposition are employed for the estimation of parameters required to implement subband LP Wiener filters. The use of subband LP Wiener filters provides advantages in terms of improved parameter estimates and also in restoring the temporal-spectral composition of speech. The method is evaluated, and compared with the parallel model combination, using the TIMIT continuous speech database with BMW and VOLVO car noise databases.

The performance of voice dialling systems often degrades rapidly as the intensity of the background noise increases. In this paper, we describe a neural network based speech enhancement technique for improving the speech recognition performance of a voice dialling system in very noisy real world type conditions. The speech samples were recorded in laboratory conditions and afterwards corrupted by adding car noise or babble noise recorded in a cafe. These noise corrupted speech samples were enhanced in cepstral domain by a context dependent multilayer perceptron (MLP) network before performing the recognition using a hidden Markov model (HMM) based speech recognition system. The accuracy of the test set increased 58%, 55% and 46% in the car noise environments having -5 dB, 0 dB and 5 dB SNRs, respectively. The accuracy of the test set increased 44%, 48% and 39% in the babble noise environments having SNR 5 dB, 10 dB and 15 dB, respectively. The accuracy remained approximately same for both car and babble noise environments when having SNR of 20 dB.