Ischemia episode detection in ECG using kernel density estimation, support vector machine and feature selection

ABSTRACT: BACKGROUND: Myocardial ischemia can be developed into more serious diseases. Early Detection of the ischemic syndrome inelectrocardiogram (ECG) more accurately and automatically can prevent it from developing into a catastrophicdisease. To this end, we propose a new method, which employs wavelets and simple feature selection. METHODS: For training and testing, the European ST-T database is used, which is comprised of 367 ischemic ST episodes in90 records. We first remove baseline wandering, and detect time positions of QRS complexes by a method basedon the discrete wavelet transform. Next, for each heart beat, we extract three features which can be used fordifferentiating ST episodes from normal: 1) the area between QRS offset and T-peak points, 2) the normalizedand signed sum from QRS offset to effective zero voltage point, and 3) the slope from QRS onset to offset point.We average the feature values for successive five beats to reduce effects of outliers. Finally we apply classifiersto those features. RESULTS: We evaluated the algorithm by kernel density estimation (KDE) and support vector machine (SVM) methods.Sensitivity and specificity for KDE were 0.939 and 0.912, respectively. The KDE classifier detects 349 ischemicST episodes out of total 367 ST episodes. Sensitivity and specificity of SVM were 0.941 and 0.923, respectively.The SVM classifier detects 355 ischemic ST episodes. CONCLUSIONS: We proposed a new method for detecting ischemia in ECG. It contains signal processing techniques of removingbaseline wandering and detecting time positions of QRS complexes by discrete wavelet transform, and featureextraction from morphology of ECG waveforms explicitly. It was shown that the number of selected featureswere sufficient to discriminate ischemic ST episodes from the normal ones. We also showed how the proposedKDE classifier can automatically select kernel bandwidths, meaning that the algorithm does not require anynumerical values of the parameters to be supplied in advance. In the case of the SVM classifier, one has to selecta single parameter.     

RsfA (YbeB) proteins are conserved ribosomal silencing factors

The YbeB (DUF143) family of uncharacterized proteins is encoded by almost all bacterial and eukaryotic genomes but not archaea. While they have been shown to be associated with ribosomes, their molecular function remains unclear. Here we show that YbeB is a ribosomal silencing factor (RsfA) in the stationary growth phase and during the transition from rich to poor media. A knock-out of the rsfA gene shows two strong phenotypes: (i) the viability of the mutant cells are sharply impaired during stationary phase (as shown by viability competition assays), and (ii) during transition from rich to poor media the mutant cells adapt slowly and show a growth block of more than 10 hours (as shown by growth competition assays). RsfA silences translation by binding to the L14 protein of the large ribosomal subunit and, as a consequence, impairs subunit joining (as shown by molecular modeling, reporter gene analysis, in vitro translation assays, and sucrose gradient analysis). This particular interaction is conserved in all species tested, including Escherichia coli, Treponema pallidum, Streptococcus pneumoniae, Synechocystis PCC 6803, as well as human mitochondria and maize chloroplasts (as demonstrated by yeast two-hybrid tests, pull-downs, and mutagenesis). RsfA is unrelated to the eukaryotic ribosomal anti-association/60S-assembly factor eIF6, which also binds to L14, and is the first such factor in bacteria and organelles. RsfA helps cells to adapt to slow-growth/stationary phase conditions by down-regulating protein synthesis, one of the most energy-consuming processes in both bacterial and eukaryotic cells.     

Proteomic analysis optimization: selective protein sample on-column retention in reverse-phase liquid chromatography

In an effort to optimize reverse-phase liquid chromatography (RPLC) for proteomics, we studied the impact of composition of the sample injection solution on protein on-column selection and retention. All the proteins studied were retained on-column when injections were made in 50% formic acid, 0.1% TFA or 8.3M urea. When formic acid was increased to 80%, the superoxide dismutase standard (MW 26,159) and 58 mouse microsomal proteins that possessed low-range molecular weights, high pIs or basic amino acid clusters were non-retained, resulting in retention selectivity during sample injection. Introducing to the 80% formic acid injection solution an organic solvent such as acetonitrile or acetonitrile-DMSO induced further retention selectivity, and increasing levels of organic solvents reduced on-column retention. The proteome was split into the proteins that were retained on-column which eluted at higher retention times (RTs), vs the proteins that collected in the injection flow-through which normally eluted at lower RTs. This protein selectivity was confirmed after fraction collection, 1D-GE and nano-LC-MS/MS. The significance of this procedure is that it can be exploited for fast extraction of small basic proteins from the bulk of the proteome and for on-column enrichment of hydrophobic proteins.     

DNA approach to solve clustering problem based on a mutual order

Clustering is regarded as a consortium of concepts and algorithms that are aimed at revealing a structure in highly dimensional data and arriving at a collection of meaningful relationships in data and information granules. The objective of this paper is to propose a DNA computing to support the development of clustering techniques. This approach is of particular interest when dealing with huge data sets, unknown number of clusters and encountering a heterogeneous character of available data. We present a detailed algorithm and show how the essential components of the clustering technique are realized through the corresponding mechanisms of DNA computing. Numerical examples offer a detailed insight into the performance of the DNA-based clustering.     

Brain spectrin exerts much stronger effect on anionic phospholipid monolayers than erythroid spectrin

Red blood cell spectrin and its nonerythroid analogues are linked to integral proteins of the membrane by several skeletal protein receptors, such as ankyrin and protein 4.1 together with p55. However, there are also many reasons for believing that they are insufficient to engender all the properties that characterise the native membrane. Therefore, we are concerned with the mechanism by which brain spectrin interacts with phospholipids of the membrane bilayer. Brain and erythrocyte spectrin were shown previously to bind phospholipid vesicles as well as monolayers prepared from aminophospholipids: phosphatidylethanolamine and phosphatidylserine and their mixtures with phosphatidylcholine (PC).In the present study, it is shown that brain spectrin binds to monolayers prepared from anionic phospholipids, such as phosphatidylinositol (PI), phosphatidic acid (PA), phosphatidyl glycerol, diphosphatidylglycerol, and their mixtures with PC. Brain spectrin injected into the subphase to reach nanomolar concentration induced a substantial increase in the surface pressure of monolayers prepared from the phospholipids and their mixtures mentioned above, possibly by penetrating them. This effect is stronger in the case of monolayers prepared from anionic phospholipids alone and weaker when monolayers were prepared from mixtures with PC. The weakest effect was observed in the case of phosphatidylinositol-4,5-bisphosphate monolayers. An interaction of brain spectrin with monolayers prepared from anionic phospholipids (PI/PC 7:3 and PA/PC 7:3) was inhibited (PI/PC much stronger than PA/PC) by purified erythrocyte ankyrin, which indicates that the binding site for those lipids is located in the β-subunit, possibly in, or in close proximity of, the ankyrin-binding site.In contrast, erythrocyte spectrin injected into the subphase induced a change in the surface pressure of monolayers prepared from anionic phospholipids, which was equal or smaller than the value of surface pressure change induced by protein without a monolayer. This effect was different from what had been observed previously for monolayers prepared from aminophospholipids and their mixtures with PC, and from the data for nonerythroid spectrin presented here.     

Characterization of compost produced from separated pig manure and a variety of bulking agents at low initial C/N ratios

The aim of this study was to investigate the composting of separated pig manure solids with or without a variety of bulking agents at a low initial C/N ratio (12.5-23.3). Compost stability was investigated using an oxygen uptake rate (OUR) test and compost maturity was investigated using a germination index test. All treatments showed typical patterns of compost temperature. Temperatures above 60 °C were achieved by Day 2, followed by a thermophilic phase (50-60 °C), which lasted for 1 to 2 weeks followed by a cooling phase. The stability of one of treatments which did not contain any bulking agent - OUR of 25 mmol O₂ kg⁻¹ OM hour⁻¹ - was negatively affected by its initial high water content (69%). The addition of a bulking agent and initial water content below 60% were necessary to compost the separated solid fraction of pig manure at a low initial C/N ratio.     

L2′ loop is critical for caspase‐7 active site formation

The active sites of caspases are composed of four mobile loops. A loop (L2) from one half of the dimer interacts with a loop (L2′) from the other half of the dimer to bind substrate. In an inactive form, the two L2′ loops form a cross‐dimer hydrogen‐bond network over the dimer interface. Although the L2′ loop has been implicated as playing a central role in the formation of the active‐site loop bundle, its precise role in catalysis has not been shown. A detailed understanding of the active and inactive conformations is essential to control the caspase function. We have interrogated the contributions of the residues in the L2′ loop to catalytic function and enzyme stability. In wild‐type and all mutants, active‐site binding results in substantial stabilization of the complex. One mutation, P214A, is significantly destabilized in the ligand‐free conformation, but is as stable as wild type when bound to substrate, indicating that caspase‐7 rests in different conformations in the absence and presence of substrate. Residues K212 and I213 in the L2′ loop are shown to be essential for substrate‐binding and thus proper catalytic function of the caspase. In the crystal structure of I213A, the void created by side‐chain deletion is compensated for by rearrangement of tyrosine 211 to fill the void, suggesting that the requirements of substrate‐binding are sufficiently strong to induce the active conformation. Thus, although the L2′ loop makes no direct contacts with substrate, it is essential for buttressing the substrate‐binding groove and is central to native catalytic efficiency.     

The CDKN2a common variants: 148 Ala/Thr and 500 C/G in 3' UTR, and their association with clinical course of melanoma

Changes in CDKN2a gene are known to be linked with sporadic melanoma and hereditary predisposition to this cancer. In the Polish population mutations in the coding region of the CDKN2a gene are rather rare, therefore the attention has been focused on polymorphisms and alterations in uncoding regions such as 3' UTR. The aim of this study was to analyze two common polymorphisms, Ala148Thr and 500 C/G, and correlate them with the clinical course of melanoma. DNA from 285 patients was analyzed and found polymorphisms were correlated with the clinical parameters employing statistical methods. The obtained results allow us to conclude: (i) survival times of 500 C/G carriers vs. cumulating proportion surviving was not statistically significant; (ii) CDKN2a polymorphism 500 C/G correlated with Ala148Thr; (iii) no correlation was observed between the 500 C/G polymorphism and age of diagnosis, localization of primary melanoma and survival time; (iv) we did not find correlation between 500 C/G and type of cancer in the family; (v) changes in the CDKN2a gene were not found in patients with second cancer.     

Electrocardiogram data mining based on frame classification by dynamic time warping matching

This paper presents an electrocardiogram (ECG) data mining scheme based on the ECG frame classification realised by a dynamic time warping (DTW) matching technique, which has been used successfully in speech recognition. We use the DTW to classify ECG frames because ECG and speech signals have similar non-stationary characteristics. The DTW mapping function is obtained by searching the frame from its end to start. A threshold is setup for DTW matching residual either to classify an ECG frame or to add a new class. Classification and establishment of a template set are carried out simultaneously. A frame is classified into a category with a minimal residual and satisfying a threshold requirement. A classification residual of 1.33% is achieved by the DTW for a 10-min ECG recording.