Supplementary material : large scale read classification for next generation sequencing

Next Generation Sequencing (NGS) has revolutionised molecular biology, resulting in an explosion of data sets and an increasing role in clinical practice. Such applications necessarily require rapid identification of the organism as a prelude to annotation and further analysis. NGS data consist of a substantial number of short sequence reads, given context through downstream assembly and annotation, a process requiring reads consistent with the assumed species or species group. Highly accurate results have been obtained for restricted sets using SVM classifiers, but such methods are difficult to parallelise and success depends on careful attention to feature selection. This work examines the problem at very large scale, using a mix of synthetic and real data with a view to determining the overall structure of the problem and the effectiveness of parallel ensembles of simpler classifiers (principally random forests) in addressing the challenges of large scale genomics.

Isolation and characterization of charge-tagged phenylperoxyl radicals in the gas phase : direct evidence for products and pathways in low temperature benzene oxidation

The phenylperoxyl radical has long been accepted as a critical intermediate in the oxidation of benzene and an archetype for arylperoxyl radicals in combustion and atmospheric chemistry. Despite being central to many contemporary mechanisms underpinning these chemistries, reports of the direct detection or isolation of phenylperoxyl radicals are rare and there is little experimental evidence connecting this intermediate with expected product channels. We have prepared and isolated two charge-tagged phenyl radical models in the gas phase [i.e., 4-(N,N,N-trimethylammonium) phenyl radical cation and 4-carboxylatophenyl radical anion] and observed their reactions with dioxygen by ion-trap mass spectrometry. Measured reaction rates show good agreement with prior reports for the neutral system (k(2)[(Me3N+)C6H4 center dot + O-2] = 2.8 x 10(-11) cm(3) molecule(-1) s(-1), Phi = 4.9%; k(2)[(-O2C)C6H4 center dot + O-2] = 5.4 x 10(-1)1 cm(3) molecule(-1) s(-1), Phi = 9.2%) and the resulting mass spectra provide unequivocal evidence for the formation of phenylperoxyl radicals. Collisional activation of isolated phenylperoxyl radicals reveals unimolecular decomposition by three pathways: (i) loss of dioxygen to reform the initial phenyl radical; (ii) loss of atomic oxygen yielding a phenoxyl radical; and (iii) ejection of the formyl radical to give cyclopentadienone. Stable isotope labeling confirms these assignments. Quantum chemical calculations for both charge-tagged and neutral phenylperoxyl radicals confirm that loss of formyl radical is accessible both thermodynamically and entropically and competitive with direct loss of both hydrogen atom and carbon dioxide.

Binary image steganographic techniques classification based on multi-class steganalysis

In this paper, we propose a new multi-class steganalysis for binary image. The proposed method can identify the type of steganographic technique used by examining on the given binary image. In addition, our proposed method is also capable of differentiating an image with hidden message from the one without hidden message. In order to do that, we will extract some features from the binary image. The feature extraction method used is a combination of the method extended from our previous work and some new methods proposed in this paper. Based on the extracted feature sets, we construct our multi-class steganalysis from the SVM classifier. We also present the empirical works to demonstrate that the proposed method can effectively identify five different types of steganography.

Automated classification of limb fractures from free-text radiology reports using a clinician-informed gazetteer methodology

Background Timely diagnosis and reporting of patient symptoms in hospital emergency departments (ED) is a critical component of health services delivery. However, due to dispersed information resources and a vast amount of manual processing of unstructured information, accurate point-of-care diagnosis is often difficult. Aims The aim of this research is to report initial experimental evaluation of a clinician-informed automated method for the issue of initial misdiagnoses associated with delayed receipt of unstructured radiology reports. Method A method was developed that resembles clinical reasoning for identifying limb abnormalities. The method consists of a gazetteer of keywords related to radiological findings; the method classifies an X-ray report as abnormal if it contains evidence contained in the gazetteer. A set of 99 narrative reports of radiological findings was sourced from a tertiary hospital. Reports were manually assessed by two clinicians and discrepancies were validated by a third expert ED clinician; the final manual classification generated by the expert ED clinician was used as ground truth to empirically evaluate the approach. Results The automated method that attempts to individuate limb abnormalities by searching for keywords expressed by clinicians achieved an F-measure of 0.80 and an accuracy of 0.80. Conclusion While the automated clinician-driven method achieved promising performances, a number of avenues for improvement were identified using advanced natural language processing (NLP) and machine learning techniques.

Classification of cancer-related death certificates using machine learning

Background Cancer monitoring and prevention relies on the critical aspect of timely notification of cancer cases. However, the abstraction and classification of cancer from the free-text of pathology reports and other relevant documents, such as death certificates, exist as complex and time-consuming activities. Aims In this paper, approaches for the automatic detection of notifiable cancer cases as the cause of death from free-text death certificates supplied to Cancer Registries are investigated. Method A number of machine learning classifiers were studied. Features were extracted using natural language techniques and the Medtex toolkit. The numerous features encompassed stemmed words, bi-grams, and concepts from the SNOMED CT medical terminology. The baseline consisted of a keyword spotter using keywords extracted from the long description of ICD-10 cancer related codes. Results Death certificates with notifiable cancer listed as the cause of death can be effectively identified with the methods studied in this paper. A Support Vector Machine (SVM) classifier achieved best performance with an overall F-measure of 0.9866 when evaluated on a set of 5,000 free-text death certificates using the token stem feature set. The SNOMED CT concept plus token stem feature set reached the lowest variance (0.0032) and false negative rate (0.0297) while achieving an F-measure of 0.9864. The SVM classifier accounts for the first 18 of the top 40 evaluated runs, and entails the most robust classifier with a variance of 0.001141, half the variance of the other classifiers. Conclusion The selection of features significantly produced the most influences on the performance of the classifiers, although the type of classifier employed also affects performance. In contrast, the feature weighting schema created a negligible effect on performance. Specifically, it is found that stemmed tokens with or without SNOMED CT concepts create the most effective feature when combined with an SVM classifier.

Automatic classification of human epithelial type 2 cell indirect immunofluorescence images using cell pyramid matching

This paper describes a novel system for automatic classification of images obtained from Anti-Nuclear Antibody (ANA) pathology tests on Human Epithelial type 2 (HEp-2) cells using the Indirect Immunofluorescence (IIF) protocol. The IIF protocol on HEp-2 cells has been the hallmark method to identify the presence of ANAs, due to its high sensitivity and the large range of antigens that can be detected. However, it suffers from numerous shortcomings, such as being subjective as well as time and labour intensive. Computer Aided Diagnostic (CAD) systems have been developed to address these problems, which automatically classify a HEp-2 cell image into one of its known patterns (eg. speckled, homogeneous). Most of the existing CAD systems use handpicked features to represent a HEp-2 cell image, which may only work in limited scenarios. We propose a novel automatic cell image classification method termed Cell Pyramid Matching (CPM), which is comprised of regional histograms of visual words coupled with the Multiple Kernel Learning framework. We present a study of several variations of generating histograms and show the efficacy of the system on two publicly available datasets: the ICPR HEp-2 cell classification contest dataset and the SNPHEp-2 dataset.

The collision induced loss of carbon monoxide from deprotonated benzyl benzoate in the gas phase. An anionic 1,2-Wittig type rearrangement

The ion PhCO2--CHPh, upon collision activation, undergoes competitive losses of CO and CO2 of which the former process produces the base peak of the spectrum. Product ion and substituent effect (Hammett) studies indicate that PhCO2--CHPh cyclises to a deprotonated hydroxydiphenyloxirane which ring opens to PhCOCH(O-)Ph. This anion then undergoes an anionic 1,2-Wittig type rearrangement {through [PhCO- (PhCHO)]} to form Ph2CHO- and CO. The mechanism of the 1,2-rearrangement has been probed by an ab initio study [at MP4(SDTQ)/6-31++G(d,p) level] of the model system HCOCH2O- →; MeO- + CO The analogous system RCO2--CHPh (R = alkyl) similarly loses CO, and the migratory aptitudes of the alkyl R groups in this reaction are Bu′ > Me > Et ∼Pri). This trend correlates with the order of anion basicities (i.e. the order of ΔG○acid values of RH), supporting the operation of an anion migration process. The loss of CO2 from PhCO2--CHPh yields Ph2CH- as the anionic product: several mechanistic scenarios are possible, one of which involves an initial ipso nucleophilic substitution.

The loss of CO from the ortho, meta and para forms of deprotonated methyl benzoate in the gas phase

The ortho, meta and para anions of methyl benzoate may be made in the source of a mass spectrometer by the S(N)2(Si) reactions between HO- and methyl (o-, m-, and p-trimethylsilyl)benzoate respectively. All three anions lose CO upon collisional activation to form the ortho anion of anisole in the ratio ortho>>meta > para. The rearrangement process is charge directed through the ortho anion. Theoretical calculations at the B3LYP/6-311++G(d,p)//HF/6-31+G(d) level of theory indicate that the conversion of the meta and para anions to the ortho anion prior to loss of CO involve 1,2-H transfer(s), rather than carbon scrambling of the methoxycarbonylphenyl anion. There are two mechanisms which can account for this rearrangement, viz. (A) cyclisation of the ortho anion centre to the carbonyl group of the ester to give a cyclic carbonyl system in which the incipient methoxide anion substitutes at one of the two equivalent ring carbons of the three membered ring to yield an intermediate which loses CO to give the ortho anion of anisole, and (B) an elimination reaction to give an intermediate benzyne-methoxycarbonyl anion complex in which the MeOCO- species acts as a MeO- donor, which then adds to benzyne to yield the ortho anion of anisole. Calculations at the B3LYP/6-311++G(d,p)//HF/6-31+G(d) level of theory indicate that (i) the barrier in the first step (the rate determining step) of process A is 87 kJ mol(-1) less than that for the synchronous benzyne process B, and (ii) there are more low frequency vibrations in the transition state for benzyne process B than for the corresponding transition state for process A. Stepwise process A has the lower barrier for the rate determining step, and the lower Arrhenius factor: we cannot differentiate between these two mechanisms on available evidence.

Syntheses of NCN and NC3N from ionic precursors in the gas phase and an unusual rearrangement of neutral NC3N : A joint experimental and theoretical study

Neutral NCN is made in a mass spectrometer by charge stripping of NCN-., while neutral dicyanocarbene NCCCN can be formed by neutralization of either the corresponding anionic and cationic species, NCCCN-. and NCCCN+.. Theoretical calculations at the RCCSD(T)/aug-cc-pVTZ//B3LYP/6-31+G(d) level of theory indicate that the (3)Sigma (-)(g) State of NCCCN is 18 kcal mol(-1) more stable than the (1)A(1) state. While the majority of neutrals formed from either NCCCN-. or NCCCN+. correspond to NCCCN, a proportion of the neutral NCCCN molecules have sufficient excess energy to effect rearrangement, as evidenced by a loss of atomic carbon in the neutralization reionization (NR) spectra of either NCCCN+. and NCCCN-.. C-13 labeling studies indicate that loss of carbon occurs statistically following or accompanied by scrambling of all three carbon atoms. A theoretical study at the B3LYP/6-31+G(d)//B3LYP/6-31+G(d) level of theory indicates that C loss is a consequence of the rearrangement sequence NCCCN --> CNCCN --> CNCNC and that C scrambling occurs within singlet CNCCN via the intermediacy of a four-membered C-2v-symmetrical transition structure.

Improved image set classification via joint sparse approximated nearest subspaces

Existing multi-model approaches for image set classification extract local models by clustering each image set individually only once, with fixed clusters used for matching with other image sets. However, this may result in the two closest clusters to represent different characteristics of an object, due to different undesirable environmental conditions (such as variations in illumination and pose). To address this problem, we propose to constrain the clustering of each query image set by forcing the clusters to have resemblance to the clusters in the gallery image sets. We first define a Frobenius norm distance between subspaces over Grassmann manifolds based on reconstruction error. We then extract local linear subspaces from a gallery image set via sparse representation. For each local linear subspace, we adaptively construct the corresponding closest subspace from the samples of a probe image set by joint sparse representation. We show that by minimising the sparse representation reconstruction error, we approach the nearest point on a Grassmann manifold. Experiments on Honda, ETH-80 and Cambridge-Gesture datasets show that the proposed method consistently outperforms several other recent techniques, such as Affine Hull based Image Set Distance (AHISD), Sparse Approximated Nearest Points (SANP) and Manifold Discriminant Analysis (MDA).

Loss of epithelial markers and acquisition of vimentin expression in adriamycin- and vinblastine-resistant human breast cancer cell lines

We have previously observed that breast cancer cell lines could exhibit either epithelial or fibroblastic phenotypes as reflected by their morphologies and intermediate filament protein expression (C. L. Sommers, D. Walker-Jones, S. E. Heckford, P. Worland, E. Valverius, R. Clark, M. Stampfer, and E. P. Gelmann, Cancer Res., 49: 4258-4263, 1989). Fibroblastoid, vimentin-expressing breast cancer cell lines are more invasive in vitro and in vivo (E. W. Thompson, S. Paik, N. Brunner, C. L. Sommers, G. Zugmaier, R. Clarke, T. B. Shima, J. Torri, S. Donahue, M. E. Lippman, G. R. Martin, and R. B. Dickson, J. Cell. Physiol., 150: 534-544, 1992). We hypothesized that a breast cancer cell with an epithelial phenotype could undergo a transition to a fibroblastic phenotype, possibly resulting in more invasive capacity. We now show that two Adriamycin-resistant MCF-7 cell lines and a vinblastine-resistant ZR-75-B cell line have undergone such a transition. Adriamycin-resistant MCF-7 cells express vimentin, have diminished keratin 19 expression, have lost cell adhesion molecule uvomorulin expression, and have reduced formation of desmosomes and tight junctions as determined by reduced immunodetection of their components desmoplakins I and II and zonula occludens (ZO)-1. Other MCF-7 cell lines selected for resistance to vinblastine and to Adriamycin and verapamil did not have these characteristics, indicating that drug selection does not invariably cause these phenotypic changes. In addition, to determine if vimentin expression in MCF-7 cells alone could manifest a fibroblastic phenotype, we transfected the full-length human vimentin complementary DNA into MCF-7 cells. Although vimentin expression was achieved in MCF-7 cells, it did not affect the phenotype of the cells in terms of the distribution of keratins, desmoplakins I and II, ZO-1, or uvomorulin or in terms of in vitro invasiveness. We conclude that vimentin expression is a marker for a fibroblastic and invasive phenotype in breast cancer cells but does not by itself give rise to this phenotype.

Encouragement of physical activity for weight loss may lead to negative psychological outcomes among girls

PURPOSE To examine correlates and consequences of parents' encouragement of girls' physical activity (PA) for weight loss (ENCLOSS). METHODS Data were collected for 181 girls, mothers and fathers when girls were 9, 11, and 13 years old. Mothers and fathers completed a self-report questionnaire of ENCLOSS (e.g., “I have talked to my daughter about how to exercise to lose weight”). Correlates of ENCLOSS that were assessed include girls' Body Mass Index (BMI) z-score and parents' modeling of and logistic support for PA. Dependent variables assessed at age 13 include girls' self-reported and objectively-measured PA, enjoyment of physical activity, and weight concerns. Associations between ENCLOSS, girls' BMI, and parent's support for PA were assessed using spearman rank correlations. To examine links between ENCLOSS and the outcome variables, scores for ENCLOSS were divided into tertiles at each age. Three groups were created including girls who were in the highest tertile at each age (high ENCLOSS), girls who were in the lowest tertile at each age (low ENCLOSS), and girls who varied in their tertile ranking (mid ENCLOSS). Group differences in the outcome variables were assessed using regression analysis (referent group: low ENCLOSS), controlling for girls' BMI and the outcome variable at age 9. RESULTS Girls' with higher BMI had mothers and fathers who reported higher ENCLOSS (r = .61-. 69, p<. 0001). Parents'reports of ENCLOSS were not associated with modeling of or logistic support for PA. Girls in the high ENCLOSS group reported significantly lower enjoyment of PA and higher weight concerns at age 13, independent of covariates. No differences in PA were noted. CONCLUSION Parents who encourage their daughters to be active for weight loss do not model PA or facilitate girls' PA. Persistent encouragement of PA for weight loss may lead to low enjoyment of PA and higher weight concerns among adolescent girls.

Reactive stepping behaviour in response to forward loss of balance predicts future falls in community-dwelling older adults

Background: a fall occurs when an individual experiences a loss of balance from which they are unable to recover. Assessment of balance recovery ability in older adults may therefore help to identify individuals at risk of falls. The purpose of this 12-month prospective study was to assess whether the ability to recover from a forward loss of balance with a single step across a range of lean magnitudes was predictive of falls. Methods: two hundred and one community-dwelling older adults, aged 65–90 years, underwent baseline testing of sensorimotor function and balance recovery ability followed by 12-month prospective falls evaluation. Balance recovery ability was defined by whether participants required either single or multiple steps to recover from forward loss of balance from three lean magnitudes, as well as the maximum lean magnitude participants could recover from with a single step. Results: forty-four (22%) participants experienced one or more falls during the follow-up period. Maximal recoverable lean magnitude and use of multiple steps to recover at the 15% body weight (BW) and 25%BW lean magnitudes significantly predicted a future fall (odds ratios 1.08–1.26). The Physiological Profile Assessment, an established tool that assesses variety of sensori-motor aspects of falls risk, was also predictive of falls (Odds ratios 1.22 and 1.27, respectively), whereas age, sex, postural sway and timed up and go were not predictive. Conclusion: reactive stepping behaviour in response to forward loss of balance and physiological profile assessment are independent predictors of a future fall in community-dwelling older adults. Exercise interventions designed to improve reactive stepping behaviour may protect against future falls.

Polarization of forecast densities : a new approach to time series classification

Time series classification has been extensively explored in many fields of study. Most methods are based on the historical or current information extracted from data. However, if interest is in a specific future time period, methods that directly relate to forecasts of time series are much more appropriate. An approach to time series classification is proposed based on a polarization measure of forecast densities of time series. By fitting autoregressive models, forecast replicates of each time series are obtained via the bias-corrected bootstrap, and a stationarity correction is considered when necessary. Kernel estimators are then employed to approximate forecast densities, and discrepancies of forecast densities of pairs of time series are estimated by a polarization measure, which evaluates the extent to which two densities overlap. Following the distributional properties of the polarization measure, a discriminant rule and a clustering method are proposed to conduct the supervised and unsupervised classification, respectively. The proposed methodology is applied to both simulated and real data sets, and the results show desirable properties.