Supervised Data Analysis Research Articles

Biodynamic, organic and integrated agriculture effects on cv. Italia table grapes juice, over a 3-year period experiment: an 1H NMR spectroscopy-based metabolomics study

BackgroundThe new trend demanding for “natural” agri-food products has encouraged the application of more sustainable and eco-friendly farming methods, which limit or avoid the use of synthetic chemicals. This approach is increasing in viticulture, one of the sectors with the highest commercial value since grapes and derived products are largely consumed foodstuffs, with appreciated nutritional and sensory features. In this work, 1H Nuclear Magnetic Resonance spectroscopy (1H NMR) was applied for the metabolic profiling of cv. Italia table grapes samples, from the same origin area, cultivated with different treatments (biodynamic, organic and integrated) and collected in three subsequent vintages. Multivariate statistical analysis was performed on NMR-data with the aim of comprehensively researching the possible influences on metabolites due to the use of diverse agricultural practices.ResultsBoth inter-annual variability (2020, 2021 and 2022 vintages) and different vineyard treatments (biodynamic, organic and integrated) resulted as significant drivers for samples differentiation in the preliminary unsupervised analysis of the (1H NMR spectra derived) metabolic profile data. Nevertheless, supervised data analyses showed that inter-vineyards variability, due to application of diverse farming methods, had a comparable discriminating effect with respect to harvesting years. Ethanol, sugars (as α-/β-glucose), organic acids (as malate) and amino acids (as arginine, leucine, glutamine) resulted the most viticultural practices-dependent metabolites. Interestingly, results from pairwise comparisons between treatments indicated the biodynamic samples with respect to the organic ones as the best-observed differentiation. This was followed by the biodynamic vs integrated and organic vs integrated samples comparisons, in decreasing discrimination order, as confirmed by the descriptiveness and predictive ability parameters of the corresponding pairwise OPLS-DA models.ConclusionsResults highlighted that metabolites’ composition in cv. Italia table grapes juice is significantly affected by the use of different kinds of vineyard managements (biodynamic, organic and integrated, here investigated). Metabolomics study, here employing 1H NMR spectroscopy combined with multivariate statistical analysis, offers powerful tools to elucidate the metabolic differences among classes of samples.Graphical

Sequence Control of the Self-Assembly of Elastin-Like Polypeptides into Hydrogels with Bespoke Viscoelastic and Structural Properties.

The biofabrication of structural proteins with controllable properties via amino acid sequence design is interesting for biomedicine and biotechnology, yet a complete framework that connects amino acid sequence to material properties is unavailable, despite great progress to establish design rules for synthesizing peptides and proteins with specific conformations (e.g., unfolded, helical, β-sheets, or β-turns) and intermolecular interactions (e.g., amphipathic peptides or hydrophobic domains). Molecular dynamics (MD) simulations can help in developing such a framework, but the lack of a standardized way of interpreting the outcome of these simulations hinders their predictive value for the design of de novo structural proteins. To address this, we developed a model that unambiguously classifies a library of de novo elastin-like polypeptides (ELPs) with varying numbers and locations of hydrophobic/hydrophilic and physical/chemical-cross-linking blocks according to their thermoresponsiveness at physiological temperature. Our approach does not require long simulation times or advanced sampling methods. Instead, we apply (un)supervised data analysis methods to a data set of molecular properties from relatively short MD simulations (150 ns). We also experimentally investigate hydrogels of those ELPs from the library predicted to be thermoresponsive, revealing several handles to tune their mechanical and structural properties: chain hydrophilicity/hydrophobicity or block distribution control the viscoelasticity and thermoresponsiveness, whereas ELP concentration defines the network permeability. Our findings provide an avenue to accelerate the design of de novo ELPs with bespoke phase behavior and material properties.

Gc-ims-tools – A new Python package for chemometric analysis of GC–IMS data

Due to its high sensitivity and resolving power, gas chromatography ion mobility spectrometry (GC-IMS) is an emerging benchtop technique for non-target screening of complex sample materials. Given the wide range of applications, such as food authenticity, custom data analysis workflows are needed. As a common basis, they necessarily share many functionalities such as file input/output, preprocessing methods, exploratory or supervised analysis and visualizations. This study introduces a new open source, fully customizable Python package for handling and analysis of GC-IMS data. A workflow to classify olive oils by geographical origin exemplarily demonstrates functionality and ease of use. Key preprocessing steps, exploratory – and supervised data analysis and feature selections are visualized.All code and detailed documentation are freely available as open source under the BSD 3-clause license at https://github.com/Charisma-Mannheim/gc-ims-tools.

HPLC-MS/MS method applied to an untargeted metabolomics approach for the diagnosis of \u201colive quick decline syndrome\u201d

Olive quick decline syndrome (OQDS) is a disorder associated with bacterial infections caused by Xylella fastidiosa subsp. pauca ST53 in olive trees. Metabolic profile changes occurring in infected olive trees are still poorly investigated, but have the potential to unravel reliable biomarkers to be exploited for early diagnosis of infections. In this study, an untargeted metabolomic method using high-performance liquid chromatography coupled to quadrupole-time-of-flight high-resolution mass spectrometry (HPLC-ESI-Q-TOF-MS) was used to detect differences in samples (leaves) from healthy (Ctrl) and infected (Xf) olive trees. Both unsupervised and supervised data analysis clearly differentiated the groups. Different metabolites have been identified as potential specific biomarkers, and their characterization strongly suggests that metabolism of flavonoids and long-chain fatty acids is perturbed in Xf samples. In particular, a decrease in the defence capabilities of the host after Xf infection is proposed because of a significant dysregulation of some metabolites belonging to flavonoid family. Moreover, oleic acid is confirmed as a putative diffusible signal factor (DSF). This study provides new insights into the host-pathogen interactions and confirms LC-HRMS-based metabolomics as a powerful approach for disease-associated biomarkers discovery in plants.Graphical abstract

Exploring blood alterations in chronic kidney disease and haemodialysis using metabolomics

Chronic kidney disease (CKD) is characterized by retention of uremic solutes. Compared to patients with non-dialysis dependent CKD, those requiring haemodialysis (HD) have increased morbidity and mortality. We wished to characterise metabolic patterns in CKD compared to HD patients using metabolomics. Prevalent non-HD CKD KDIGO stage 3b–4 and stage 5 HD outpatients were screened at a single tertiary hospital. Various liquid chromatography approaches hyphenated with mass spectrometry were used to identify 278 metabolites. Unsupervised and supervised data analyses were conducted to characterize metabolic patterns. 69 patients were included in the CKD group and 35 in the HD group. Unsupervised data analysis showed clear clustering of CKD, pre-dialysis (preHD) and post-dialysis (postHD) patients. Supervised data analysis revealed qualitative as well as quantitative differences in individual metabolites profiles between CKD, preHD and postHD states. An original metabolomics framework could discriminate between CKD stages and highlight HD effect based on 278 identified metabolites. Significant differences in metabolic patterns between CKD and HD patients were found overall as well as for specific metabolites. Those findings could explain clinical discrepancies between patients requiring HD and those with earlier stage of CKD.

ENose breath prints as a surrogate biomarker for classifying patients with asthma by atopy

Electronic noses (eNoses) are emerging point-of-care tools that may help in the subphenotyping of chronic respiratory diseases such as asthma. We aimed to investigate whether eNoses can classify atopy in pediatric and adult patients with asthma. Participants with asthma and/or wheezing from 4 independent cohorts were included; BreathCloud participants (n= 429), Unbiased Biomarkers in Prediction of Respiratory Disease Outcomes adults (n= 96), Unbiased Biomarkers in Prediction of Respiratory Disease Outcomes pediatric participants (n= 100), and Pharmacogenetics of Asthma Medication in Children: Medication with Anti-Inflammatory Effects 2 participants (n= 30). Atopy was defined as a positive skin prick test result (≥3 mm) and/or a positive specific IgE level (≥0.35 kU/L) for common allergens. Exhaled breath profiles were measured by using either an integrated eNose platform or the SpiroNose. Data were divided into 2 training and 2 validation sets according to the technology used. Supervised data analysis involved the use of 3 different machine learning algorithms to classify patients with atopic versus nonatopic asthma with reporting of areas under the receiver operating characteristic curves as a measure of model performance. In addition, an unsupervised approach was performed by using a bayesian network to reveal data-driven relationships between eNose volatile organic compound profiles and asthma characteristics. Breath profiles of 655 participants (n= 601 adults and school-aged children with asthma and 54 preschool children with wheezing [68.2% of whom were atopic]) were included in this study. Machine learning models utilizing volatile organic compound profiles discriminated between atopic and nonatopic participants with areas under the receiver operating characteristic curves of at least 0.84 and 0.72 in the training and validation sets, respectively. The unsupervised approach revealed that breath profiles classifying atopy are not confounded by other patient characteristics. eNoses accurately detect atopy in individuals with asthma and wheezing in cohorts with different age groups and could be used in asthma phenotyping.

LC-MS untargeted approach showed that methyl jasmonate application on Vitis labrusca L. grapes increases phenolics at subtropical Brazilian regions.

Vitis labrusca L. grapes are largely cultivated in Brazil, but the tropical climate negatively affects the phenols content, especially anthocyanin. According to the projections of the incoming climatic changes, the climate of several viticulture zone might change to tropical. Therefore, researches are focusing on increasing grape phenols content; with methyl jasmonate application (MeJa) is considered a good alternative. The aim was to investigate with an untargeted approach the metabolic changes caused by the MeJa pre-harvest application on two Vitis labrusca L. cultivars grapes, both of them grown in two Brazilian regions. Isabel Precoce and Concord grapes cultivated under subtropical climate, in the south and southeast of Brazil, received MeJa pre-harvest treatment. Grape metabolome was extracted and analyzed with a MS based metabolomics protocol by UPLC-HRMS-QTOF. Unsupervised data analysis revealed a clear separation between the two regions and the two cultivars, while supervised data analysis revealed biomarkers between the MeJa treatment group and the control group. Among the metabolites positively affected by MeJa were (a) flavonoids with a high degree of methylation at the B-ring (malvidin and peonidin derivatives and isorhamentin) for Isabel Precoce grapes; (b) glucosides of hydroxycinnamates, gallocatechin, epigallocatechin and cis-piceid for Concord grapes; and (c) hydroxycinnamates esters with tartaric acid, and procyanidins for the Southeast region grapes. These results suggest that MeJa can be used as elicitor to secondary metabolism in grapes grown even under subtropical climate, affecting phenolic biosynthesis.

A new method for real-time monitoring of volatiles in frying fumes using proton transfer reaction mass spectrometry with time-of-flight analyser

To safeguard the consumers’ well-being, it is necessary to develop novel methods for determination of carcinogens in food, including volatiles generated during frying. The currently used procedures for analysis of volatile fraction of vegetable oils are not based on real-time measurements and thus do not enable the determination of carcinogenic compounds in frying fumes; instead, only the headspace or liquid fraction is sampled. In this article, described is an approach in which proton transfer reaction mass spectrometry with time-of-flight analyser (PTR-TOFMS) was used for real-time monitoring of carcinogenic compounds generated during thermal degradation of rapeseed oil. Using PTR-MS, it was possible to monitor the concentration of known volatile carcinogens according to the International Agency for Research on Cancer (IARC), alongside BTEX compounds, acrolein, and selected aldehydes. Moreover, the applicability of several supervised data analysis methods for the classification of oil samples according to their degree of thermal degradation was presented, with best results obtained using the k-nearest neighbours algorithm. Proton transfer reaction mass spectrometry is a powerful technique for the determination of carcinogenic compounds generated during thermal degradation of edible oils. Further investigation of the chemical processes which occur during frying can lead to improvement of food safety.Graphical abstract

Abstract 1073: Breast cancer intravasation signature from patient fine-needle aspirates

Abstract A key step in the progression of metastatic breast cancer is the process of blood vessel intravasation. Understanding the molecular mechanisms that underlie this crucial step is essential in designing strategies to block the same with an aim to prevent metastatic disease. Recently our group has developed and published an in vitro intravasation assay (1) that recapitulates the process of intravasation and allows the use of patient derived breast cancer cells. Previously our group had identified a breast cancer invasion signature that helps facilitate clinical decision making in breast cancer patients (2). Here using fine needle aspirates from breast cancer resections we collected viable cancer cells which were labeled green using vital dyes and mixed with a macrophage cell line labeled red and layered on the top compartment of a transwell. The bottom of the transwell membrane contains a layer of extracellular matrix and a tight layer of HUVECs. Green cancer cells were separated one at a time using a florescent microscope and a micromanipulator. The cancer cells capable of intravasation are divided into two groups, one group that has crossed the entire layer and fallen to the bottom of the well, and a second group that crossed the endothelial cells but are still bound to the bottom of the endothelial layer. Cancer cells from top of the transwell which did not intravasate were also collected. RNA was extracted from the few cells collected by micromanipulation from all the three groups, amplified and converted to Cy3 labeled cDNA. The labeled cDNA was hybridized to whole genome microarrays. Gene expression was compared between the three groups of cells. Both unsupervised and supervised data analysis was performed on the three groups of cells from five patients. Significance Analysis of Microarrays, Ingenuity Pathway Analysis and Gene Set Enrichment Analysis was performed on these datasets. Supervised analysis and validation of the altered genes was performed using qRT-PCR. Preliminary data analysis indicates the upregulation of the genes associated with Breast Cancer Stem cells and DNA repair. This is the first report of whole genome gene expression analysis from cells collected by FNA from patients, separated in vitro into intravasation competent and intravasation incompetent and unsupervised whole genome analysis performed to identify intravasation specific genes and pathways. Once unique targets and pathways are identified in intrvasation competent cells, a number of additional function blocking steps will be designed and undertaken. (1) Pignatelli J., Goswami S., et. al. Invasive breast carcinoma cells from patients exhibit MenaINV- and macrophage-dependent transendothelial migration. Sci Signal. 2014 Nov 25;7(353) (2) Karagiannis GS., Goswami S., Jones JG., Oktay MH., and Condeelis JS. Signatures of breast cancer metastasis at a glance. J Cell Sci. 2016 May 1;129(9):1751-8. Citation Format: Gargi Bandyopadhyaya, Srinjoy Goswami, John S. Condeelis, Maja H. Oktay, Sumanta Goswami. Breast cancer intravasation signature from patient fine-needle aspirates [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1073.

MicroRNA expression patterns in canine mammary cancer show significant differences between metastatic and non-metastatic tumours

BackgroundMicroRNAs may act as oncogenes or tumour suppressor genes, which make these small molecules potential diagnostic/prognostic factors and targets for anticancer therapies. Several common oncogenic microRNAs have been found for canine mammary cancer and human breast cancer. On account of this, large-scale profiling of microRNA expression in canine mammary cancer seems to be important for both dogs and humans.MethodsExpression profiles of 317 microRNAs in 146 canine mammary tumours of different histological type, malignancy grade and clinical history (presence/absence of metastases) and in 25 control samples were evaluated. The profiling was performed using microarrays. Significance Analysis of Microarrays test was applied in the analysis of microarray data (both unsupervised and supervised data analyses were performed). Validation of the obtained results was performed using real-time qPCR. Subsequently, predicted targets for the microRNAs were searched for in miRBase.ResultsResults of the unsupervised analysis indicate that the primary factor separating the samples is the metastasis status. Predicted targets for microRNAs differentially expressed in the metastatic vs. non-metastatic group are mostly engaged in cell cycle regulation, cell differentiation and DNA-damage repair. On the other hand, the supervised analysis reveals clusters of differentially expressed microRNAs unique for the tumour type, malignancy grade and metastasis factor.ConclusionsThe most significant difference in microRNA expression was observed between the metastatic and non-metastatic group, which suggests a more important role of microRNAs in the metastasis process than in the malignant transformation. Moreover, the differentially expressed microRNAs constitute potential metastasis markers. However, validation of cfa-miR-144, cfa-miR-32 and cfa-miR-374a levels in blood samples did not follow changes observed in the non-metastatic and metastatic tumours.

Performance-Estimation Properties of Cross-Validation-Based Protocols with Simultaneous Hyper-Parameter Optimization

In a typical supervised data analysis task, one needs to perform the following two tasks: (a) select an optimal combination of learning methods (e.g., for variable selection and classifier) and tune their hyper-parameters (e.g., K in K-NN), also called model selection, and (b) provide an estimate of the performance of the final, reported model. Combining the two tasks is not trivial because when one selects the set of hyper-parameters that seem to provide the best estimated performance, this estimation is optimistic (biased/overfitted) due to performing multiple statistical comparisons. In this paper, we discuss the theoretical properties of performance estimation when model selection is present and we confirm that the simple Cross-Validation with model selection is indeed optimistic (overestimates performance) in small sample scenarios and should be avoided. We present in detail and investigate the theoretical properties of the Nested Cross Validation and a method by Tibshirani and Tibshirani for removing the estimation bias. In computational experiments with real datasets both protocols provide conservative estimation of performance and should be preferred. These statements hold true even if feature selection is performed as preprocessing.

Differential diagnosis of pleural mesothelioma using Logic Learning Machine.

BackgroundTumour markers are standard tools for the differential diagnosis of cancer. However, the occurrence of nonspecific symptoms and different malignancies involving the same cancer site may lead to a high proportion of misclassifications.Classification accuracy can be improved by combining information from different markers using standard data mining techniques, like Decision Tree (DT), Artificial Neural Network (ANN), and k-Nearest Neighbour (KNN) classifier. Unfortunately, each method suffers from some unavoidable limitations. DT, in general, tends to show a low classification performance, whereas ANN and KNN produce a "black-box" classification that does not provide biological information useful for clinical purposes.MethodsLogic Learning Machine (LLM) is an innovative method of supervised data analysis capable of building classifiers described by a set of intelligible rules including simple conditions in their antecedent part. It is essentially an efficient implementation of the Switching Neural Network model and reaches excellent classification accuracy while keeping low the computational demand.LLM was applied to data from a consecutive cohort of 169 patients admitted for diagnosis to two pulmonary departments in Northern Italy from 2009 to 2011. Patients included 52 malignant pleural mesotheliomas (MPM), 62 pleural metastases (MTX) from other tumours and 55 benign diseases (BD) associated with pleurisies. Concentration of three tumour markers (CEA, CYFRA 21-1 and SMRP) was measured in the pleural fluid of each patient and a cytological examination was also carried out.The performance of LLM and that of three competing methods (DT, KNN and ANN) was assessed by leave-one-out cross-validation.ResultsLLM outperformed all other considered methods. Global accuracy was 77.5% for LLM, 72.8% for DT, 54.4% for KNN, and 63.9% for ANN, respectively. In more details, LLM correctly classified 79% of MPM, 66% of MTX and 89% of BD. The corresponding figures for DT were: MPM = 83%, MTX = 55% and BD = 84%; for KNN: MPM = 58%, MTX = 45%, BD = 62%; for ANN: MPM = 71%, MTX = 47%, BD = 76%.Finally, LLM provided classification rules in a very good agreement with a priori knowledge about the biological role of the considered tumour markers.ConclusionsLLM is a new flexible tool potentially useful for the differential diagnosis of pleural mesothelioma.

Evaluation of reverse phase protein array (RPPA)-based pathway-activation profiling in 84 non-small cell lung cancer (NSCLC) cell lines as platform for cancer proteomics and biomarker discovery

The reverse phase protein array (RPPA) approach was employed for a quantitative analysis of 71 cancer-relevant proteins and phosphoproteins in 84 non-small cell lung cancer (NSCLC) cell lines and by monitoring the activation state of selected receptor tyrosine kinases, PI3K/AKT and MEK/ERK1/2 signaling, cell cycle control, apoptosis, and DNA damage. Additional information on NSCLC cell lines such as that of transcriptomic data, genomic aberrations, and drug sensitivity was analyzed in the context of proteomic data using supervised and non-supervised approaches for data analysis. First, the unsupervised analysis of proteomic data indicated that proteins clustering closely together reflect well-known signaling modules, e.g. PI3K/AKT- and RAS/RAF/ERK-signaling, cell cycle regulation, and apoptosis. However, mutations of EGFR, ERBB2, RAF, RAS, TP53, and PI3K were found dispersed across different signaling pathway clusters. Merely cell lines with an amplification of EGFR and/or ERBB2 clustered closely together on the proteomic, but not on the transcriptomic level. Secondly, supervised data analysis revealed that sensitivity towards anti-EGFR drugs generally correlated better with high level EGFR phosphorylation than with EGFR abundance itself. High level phosphorylation of RB and high abundance of AURKA were identified as candidates that can potentially predict sensitivity towards the aurora kinase inhibitor VX680. Examples shown demonstrate that the RPPA approach presents a useful platform for targeted proteomics with high potential for biomarker discovery. This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge.

Decitabine Response Associated Gene Expression Patterns In Acute Myeloid Leukemia (AML)

The hypomethylating agent decitabine (DAC) represents a therapeutic option for acute myeloid leukemia (AML) patients who are not eligible for an intensive treatment regime. However, there are no biomarkers available yet that can predict patients who will likely benefit from this epigenetic therapy.Therefore, we executed a gene expression analysis prior to the treatment of patients with DAC in order to evaluate gene expression patterns associated with response to DAC that ultimately might be used to predict DAC outcome. Patients had been entered in a multicenter phase II trial of DAC as first-line treatment of older AML patients judged unfit for induction chemotherapy (Lübbert et al. Haematologica 2011; NCT00866073).Gene expression was profiled in selected DAC responders (n=17) and non-responders (n=19; non-response was defined as stable disease or progressive disease). These groups did not show significant differences regarding age, gender, performance status, blast counts and cytogenetics. Supervised data analysis strategies were applied to identify genes and gene patterns associated with DAC response.While the study cohort comprised a heterogeneous group of AML patients, a class comparison analysis nevertheless could reveal a DAC response associated gene pattern comprising 301 genes at a significance level of p<0.05. This signature was enriched for genes belonging to pathways that are essential in immune response and tumor suppressor function. Among these genes that were significantly associated with no DAC response included IFI44L, IFI27, PDK4, MX1, FAS, and ITGB2; in contrast to SLC24A3, MUM1, TNFSF9, DBN1, ABAT, and DDX52, which were significantly higher expressed in patients that showed response to DAC treatment. Significantly over-expressed in the DAC non-responder group, the immune and inflammation-related genes IFI44L and IFI27 might reflect a hyperstimulated, but insufficient immune system as has been recently shown in myelofibrosis. As DAC was shown to have the capability to induce cancer testis antigens, thereby generating an efficient immune response with tumor cell lysis by CD8+ T-lymphocytes, an impaired immune system may prevent response to DAC. Furthermore, the non-response signature contained known poor prognostic markers such as PDK4, which has been associated with EVI1 and FLT3-ITD mediated signaling. In addition, we observed high expression of MX1 and FAS in the non-response group. Notably, both genes have been shown to be repressed by promoter hypermethylation in distinct AML subtypes and DAC treatment was able to upregulate their expression levels. In contrast, high pre-treatment expression levels might indicate that in the respective AML cases deregulated promotor methylation might not be the prominent pathomechanism, and thus these cases might less likely benefit from DAC treatment. Finally, we found ITGB2, encoding for an integral cell-surface protein participating in cell-surface mediated signaling, associated with DAC resistance. As recently ITGB3, another member of this integrin protein family, was shown mandatory for leukemogenesis, but not relevant for normal hematopoiesis, high expression of ITGB2 might also play a role in AML and point to leukemias where epigenetic deregulation at the DNA level seems to be a less prone pathomechanism. Among the group of genes linked with response to DAC treatment TNFSF9 can act as cytotoxic leukemic specific T-cell inducer, which has previously been correlated with unfavorable AML subtypes and poor outcome. However, due to the immunomodulation of DAC it seems that the poor prognostic impact of TNFSF9 might be overcome by DAC, thereby rendering TNFSF9 a positive marker for DAC response. In accordance, we found that several genes of the 4-1BB-dependent immune response pathway, including TNFSF9, were more highly expressed in DAC responding patients. Finally, MUM1 encodes also a gene important for interferon dependent immune response, thereby further underscoring a potential immunomodulation effect of DAC. In summary, we were able to elucidate a gene signature which could be used to predict response to DAC treatment in AML. While this gene expression pattern included many genes involved in the immune response, thereby suggesting that the DAC treatment effect is at least in part depending on immunomodulatory effects, further studies are warranted to evaluate the respective markers in larger AML cohorts. Disclosures:Schlenk:Amgen: Research Funding; Pfizer: Research Funding; Novartis: Research Funding; Chugai: Research Funding; Ambit: Honoraria.

Differential Protein Expression Profiles Between Plasmodium falciparum Parasites Isolated From Subjects Presenting With Pregnancy-Associated Malaria and Uncomplicated Malaria in Benin

Plasmodium falciparum is responsible for severe malaria, including pregnancy-associated malaria (PAM). During intra-erythrocytic maturation, the infected erythrocyte (iE) membrane is modified by insertion of parasite-derived proteins, primarily consisting of variant surface antigens such as P. falciparum erythrocyte membrane protein-1. To identify new PAM-specific parasite membrane proteins, we conducted a mass spectrometry-based proteomic study and compared the protein expression profiles of 10 PAM and 10 uncomplicated malaria (UM) samples. We focused on the 454/1139 membrane-associated and hypothetical proteins for comparative analysis. Using filter-based feature-selection methods combined with supervised data analysis, we identified a subset of 53 proteins that distinguished PAM and UM samples. Up to 19/20 samples were correctly assigned to their respective clinical group. A hierarchical clustering analysis of these 53 proteins based on the similarity of their expression profiles revealed 2 main clusters of 40 and 13 proteins that were under- or over-expressed, respectively, in PAM. VAR2CSA is identified and associated with PAM, validating our experimental approach. Other PAM-predictive proteins included PFI1785w, PF14_0018, PFB0115w, PFF0325c, and PFA_0410w. These proteomics data demonstrate the involvement of selected proteins in the pathophysiology of PAM, providing new insights for the definition of potential new targets for a vaccine against PAM.

Nonlinear Feature Extraction in a Logarithmic Space with Evolutionary Algorithms

The current paper presents a method to deliver non- linear projections of a data set that discriminate between existing labeled groups of data items. Inspired from traditional linear Pro- jection Pursuit and Linear Discriminant Analysis, the new method seeks nonlinear combinations of attributes as polynomials that maximize Fisher's criterion. The search for the monomials in a polynomial is conducted in a logarithmic space in order to reduce computational complexity. The selection of monomials and the optimization of weights that conduct to the nonlinear projection are performed with a multi-modal Genetic Algorithm hybridized with Differential Evolution. By alleviating the drawbacks driven from the linearity assumptions in traditional Projection Pursuit, the new method could gain a wide applicability in both unsuper- vised and supervised data analysis.

Regularized partial least squares with an application to NMR spectroscopy

High-dimensional data common in genomics, proteomics, and chemometrics often contains complicated correlation structures. Recently, partial least squares (PLS) and Sparse PLS methods have gained attention in these areas as dimension reduction techniques in the context of supervised data analysis. We introduce a framework for Regularized PLS by solving a relaxation of the SIMPLS optimization problem with penalties on the PLS loadings vectors. Our approach enjoys many advantages including flexibility, general penalties, easy interpretation of results, and fast computation in high-dimensional settings. We also outline extensions of our methods leading to novel methods for non-negative PLS and generalized PLS, an adoption of PLS for structured data. We demonstrate the utility of our methods through simulations and a case study on proton Nuclear Magnetic Resonance (NMR) spectroscopy data.

Transition of highly specific microRNA expression patterns in association with discrete maturation stages of human granulopoiesis

Transition of highly specific microRNA expression patterns in association with discrete maturation stages of human granulopoiesis

Abstract 2606: Absence of transcriptomic signature of response to chemotherapy in metastatic colorectal carcinoma patients

Abstract Using whole genome tumour gene expression profiling in patients treated for metastatic colorectal cancer, we attempted to define a signature able to discriminate between responders and non-responders to first-line chemotherapy. A total of 42 patients were prospectively included between 2003 and 2006. They all suffered from histologically proven colorectal adenocarcinoma and were metastatic at the time of diagnosis. All of them gave informed consent to the study. They were treated in first line with 5-fluorouracil (according to the de Gramont regimen in most cases) associated either with irinotecan (FolFiri regimen, 180 mg/m2 every two weeks, 22 pts) or with oxaliplatin (FolFox 4 regimen, 85 mg/m2, 20 pts) according to physician's decision. Response was evaluated using WHO criteria every 2 months after initiation of chemotherapy and patients were assigned to complete response (1 pt), partial response (24 pts), stable disease (13 pts) or progressive disease (4 pts). From a tumour sample obtained at surgery (18 pts) or by biopsy (24 pts) and kept in RNA later, we extracted and purified total RNA. Gene expression profiling was performed by Cogenics using microarrays from Applied Biosystems (Human Genome Survey Microarray v2.0). We used three different methods for supervised data analysis: analysis of variance, SAM (statistical analysis of microarrays) and Limma (Bioconductor). We compared responders to non-responders in the whole set of patients and in the subsets of patients who received either irinotecan or oxaliplatin. With none of the three methods we were able to obtain a stable signature, i.e. a group of genes consistently associated with response to chemotherapy. On the contrary, we obtained very homogeneous gene expression profiles which were even difficult to clusterise using a non-supervised approach. These negative results show that the determinants of response to chemotherapy should be sought not only in the tumour molecular characteristics, but also among the processes leading to drug availability to the tumour (drug dose, drug disposition, etc.). Especially, the relatively low dose of irinotecan which is of standard use in Europe does not allow to recruit all the responders who could be identified at higher dosage, given the sharp dose-efficacy curve that has been identified long ago for this drug. Note: This abstract was not presented at the AACR 101st Annual Meeting 2010 because the presenter was unable to attend. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2606.

Time dependence of the aroma pattern emitted by an encapsulated essence studied by means of electronic noses and chemometric analysis

This work studies the changes along days of the aroma released from a flavour encapsulated in a polysaccharide gel matrix using the electronic nose methodology. The purpose is to explore the capacity of the sensor array to assign a pattern of aroma to the corresponding release day within a total period of five days. Different procedures of data treatment and analysis are compared in order to achieve the maximum of information of the system under study in conditions where the number of measurements is limited. Raw and normalized sensor signals are processed using various unsupervised and supervised data analysis algorithms such as Principal Component Analysis, Kohonen-Self Organizing Maps, Cluster Analysis, Multiple Discriminant Analysis and two types of Artificial Neural Networks (BP-ANN and RBF-ANN). Accurate assignation of the number of release days is obtained with a successful classification up to four classes associated to samples at increasing days of aroma release. The relative advantages and drawbacks of the different procedures and data manipulations are discussed.