Adi L. Tarca

A novel signaling pathway impact analysis

MOTIVATION Gene expression class comparison studies may identify hundreds or thousands of genes as differentially expressed (DE) between sample groups. Gaining biological insight from the result of such experiments can be approached, for instance, by identifying the signaling pathways impacted by the observed changes. Most of the existing pathway analysis methods focus on either the number of DE genes observed in a given pathway (enrichment analysis methods), or on the correlation between the pathway genes and the class of the samples (functional class scoring methods). Both approaches treat the pathways as simple sets of genes, disregarding the complex gene interactions that these pathways are built to describe. RESULTS We describe a novel signaling pathway impact analysis (SPIA) that combines the evidence obtained from the classical enrichment analysis with a novel type of evidence, which measures the actual perturbation on a given pathway under a given condition. A bootstrap procedure is used to assess the significance of the observed total pathway perturbation. Using simulations we show that the evidence derived from perturbations is independent of the pathway enrichment evidence. This allows us to calculate a global pathway significance P-value, which combines the enrichment and perturbation P-values. We illustrate the capabilities of the novel method on four real datasets. The results obtained on these data show that SPIA has better specificity and more sensitivity than several widely used pathway analysis methods. AVAILABILITY SPIA was implemented as an R package available at http://vortex.cs.wayne.edu/ontoexpress/

Machine Learning and Its Applications to Biology

The term machine learning refers to a set of topics dealing with the creation and evaluation of algorithms that facilitate pattern recognition, classification, and prediction, based on models derived from existing data. Two facets of mechanization should be acknowledged when considering machine learning in broad terms. Firstly, it is intended that the classification and prediction tasks can be accomplished by a suitably programmed computing machine. That is, the product of machine learning is a classifier that can be feasibly used on available hardware. Secondly, it is intended that the creation of the classifier should itself be highly mechanized, and should not involve too much human input. This second facet is inevitably vague, but the basic objective is that the use of automatic algorithm construction methods can minimize the possibility that human biases could affect the selection and performance of the algorithm. Both the creation of the algorithm and its operation to classify objects or predict events are to be based on concrete, observable data. The history of relations between biology and the field of machine learning is long and complex. An early technique [1] for machine learning called the perceptron constituted an attempt to model actual neuronal behavior, and the field of artificial neural network (ANN) design emerged from this attempt. Early work on the analysis of translation initiation sequences [2] employed the perceptron to define criteria for start sites in Escherichia coli. Further artificial neural network architectures such as the adaptive resonance theory (ART) [3] and neocognitron [4] were inspired from the organization of the visual nervous system. In the intervening years, the flexibility of machine learning techniques has grown along with mathematical frameworks for measuring their reliability, and it is natural to hope that machine learning methods will improve the efficiency of discovery and understanding in the mounting volume and complexity of biological data. This tutorial is structured in four main components. Firstly, a brief section reviews definitions and mathematical prerequisites. Secondly, the field of supervised learning is described. Thirdly, methods of unsupervised learning are reviewed. Finally, a section reviews methods and examples as implemented in the open source data analysis and visualization language R (http://www.r-project.org).

The composition and stability of the vaginal microbiota of normal pregnant women is different from that of non-pregnant women

BackgroundThis study was undertaken to characterize the vaginal microbiota throughout normal human pregnancy using sequence-based techniques. We compared the vaginal microbial composition of non-pregnant patients with a group of pregnant women who delivered at term.ResultsA retrospective case–control longitudinal study was designed and included non-pregnant women (n = 32) and pregnant women who delivered at term (38 to 42 weeks) without complications (n = 22). Serial samples of vaginal fluid were collected from both non-pregnant and pregnant patients. A 16S rRNA gene sequence-based survey was conducted using pyrosequencing to characterize the structure and stability of the vaginal microbiota. Linear mixed effects models and generalized estimating equations were used to identify the phylotypes whose relative abundance was different between the two study groups. The vaginal microbiota of normal pregnant women was different from that of non-pregnant women (higher abundance of Lactobacillus vaginalis, L. crispatus, L. gasseri and L. jensenii and lower abundance of 22 other phylotypes in pregnant women). Bacterial community state type (CST) IV-B or CST IV-A characterized by high relative abundance of species of genus Atopobium as well as the presence of Prevotella, Sneathia, Gardnerella, Ruminococcaceae, Parvimonas, Mobiluncus and other taxa previously shown to be associated with bacterial vaginosis were less frequent in normal pregnancy. The stability of the vaginal microbiota of pregnant women was higher than that of non-pregnant women; however, during normal pregnancy, bacterial communities shift almost exclusively from one CST dominated by Lactobacillus spp. to another CST dominated by Lactobacillus spp.ConclusionWe report the first longitudinal study of the vaginal microbiota in normal pregnancy. Differences in the composition and stability of the microbial community between pregnant and non-pregnant women were observed. Lactobacillus spp. were the predominant members of the microbial community in normal pregnancy. These results can serve as the basis to study the relationship between the vaginal microbiome and adverse pregnancy outcomes.

A primate subfamily of galectins expressed at the maternal–fetal interface that promote immune cell death

Galectins are proteins that regulate immune responses through the recognition of cell-surface glycans. We present evidence that 16 human galectin genes are expressed at the maternal–fetal interface and demonstrate that a cluster of 5 galectin genes on human chromosome 19 emerged during primate evolution as a result of duplication and rearrangement of genes and pseudogenes via a birth and death process primarily mediated by transposable long interspersed nuclear elements (LINEs). Genes in the cluster are found only in anthropoids, a group of primate species that differ from their strepsirrhine counterparts by having relatively large brains and long gestations. Three of the human cluster genes (LGALS13, -14, and -16) were found to be placenta-specific. Homology modeling revealed conserved three-dimensional structures of galectins in the human cluster; however, analyses of 24 newly derived and 69 publicly available sequences in 10 anthropoid species indicate functional diversification by evidence of positive selection and amino acid replacements in carbohydrate-recognition domains. Moreover, we demonstrate altered sugar-binding capacities of 6 recombinant galectins in the cluster. We show that human placenta-specific galectins are predominantly expressed by the syncytiotrophoblast, a primary site of metabolic exchange where, early during pregnancy, the fetus comes in contact with immune cells circulating in maternal blood. Because ex vivo functional assays demonstrate that placenta-specific galectins induce the apoptosis of T lymphocytes, we propose that these galectins reduce the danger of maternal immune attacks on the fetal semiallograft, presumably conferring additional immune tolerance mechanisms and in turn sustaining hemochorial placentation during the long gestation of anthropoid primates.

Villitis of unknown etiology is associated with a distinct pattern of chemokine up-regulation in the feto-maternal and placental compartments: implications for conjoint maternal allograft rejection and maternal anti-fetal graft-versus-host disease.

The co-presence of histoincompatible fetal and maternal cells is a characteristic of human placental inflammation. Villitis of unknown etiology (VUE), a destructive inflammatory lesion of villous placenta, is characterized by participation of Hofbauer cells (placental macrophages) and maternal T cells. In contrast to acute chorioamnionitis of infection-related origin, the fundamental immunopathology of VUE is unknown. This study was performed to investigate the placental transcriptome of VUE and to determine whether VUE is associated with systemic maternal and/or fetal inflammatory response(s). Comparison of the transcriptome between term placentas without and with VUE revealed differential expression of 206 genes associated with pathways related to immune response. The mRNA expression of a subset of chemokines and their receptors (CXCL9, CXCL10, CXCL11, CXCL13, CCL4, CCL5, CXCR3, CCR5) was higher in VUE placentas than in normal placentas (p < 0.05). Analysis of blood cell mRNA showed a higher expression of CXCL9 and CXCL13 in the mother, and CXCL11 and CXCL13 in the fetus of VUE cases (p < 0.05). The median concentrations of CXCL9, CXCL10, and CXCL11 in maternal and fetal plasma were higher in VUE (p < 0.05). Comparison of preterm cases without and with acute chorioamnionitis revealed elevated CXCL9, CXCL10, CXCL11, and CXCL13 concentrations in fetal plasma (p < 0.05), but not in maternal plasma with chorioamnionitis. We report for the first time the placental transcriptome of VUE. A systemic derangement of CXC chemokines in maternal and fetal circulation distinguishes VUE from acute chorioamnionitis. We propose that VUE be a unique state combining maternal allograft rejection and maternal antifetal graft-vs-host disease mechanisms.

Maternal plasma concentrations of angiogenic/anti-angiogenic factors are of prognostic value in patients presenting to the obstetrical triage area with the suspicion of preeclampsia

Objective: To determine whether maternal plasma concentrations of placental growth factor (PlGF), soluble endoglin (sEng), soluble vascular endothelial growth factor receptor-1 (sVEGFR-1) and -2 could identify patients at risk for developing preeclampsia (PE) requiring preterm delivery. Study design: Patients presenting with the diagnosis “rule out PE” to the obstetrical triage area of our hospital at <37 weeks of gestation (n=87) were included in this study. Delivery outcomes were used to classify patients into four groups: I) patients without PE or those with gestational hypertension (GHTN) or chronic hypertension (CHTN) who subsequently developed PE at term (n = 19); II): mild PE who delivered at term (n = 15); III): mild disease (mild PE, GHTN, CHTN) who subsequently developed severe PE requiring preterm delivery (n = 26); and IV): diagnosis of severe PE (n = 27). Plasma concentrations of PlGF, sEng, sVEGFR-1 and -2 were determined at the time of presentation by ELISA. Reference ranges for analytes were constructed by quantile regression in our laboratory (n = 180; 1046 samples). Comparisons among groups were performed using multiples of the median (MoM) and parametric statistics after log transformation. Receiver operating characteristic curves, logistic regression and survival analysis were employed for analysis. Results: The mean MoM plasma concentration of PlGF/sVEGFR-1, PlGF/sEng, PlGF, sVEGFR-1 and -2, and sEng in Group III was significantly different from Group II (all p < 0.05). A plasma concentration of PlGF/sVEGFR-1 ≤ 0.05 MoM or PlGF/sEng ≤0.07 MoM had the highest likelihood ratio of a positive test (8.3, 95% CI 2.8–25 and 8.6, 95% CI 2.9–25, respectively), while that of PlGF ≤0.396 MoM had the lowest likelihood ratio of a negative test (0.08, 95% CI 0.03–0.25). The association between low plasma concentrations of PlGF/sVEGFR-1 (≤0.05 MoM) as well as that of PlGF/sEng (≤0.07 MoM) and the development of severe PE remained significant after adjusting for gestational age at presentation, average systolic and diastolic blood pressure, and a history of chronic hypertension [adjusted odds ratio (OR) = 27 (95% CI 6.4–109) and adjusted OR 30 (95% CI 6.9–126), respectively]. Among patients who presented <34 weeks gestation (n = 59), a plasma concentration of PlGF/sVEGFR-1 < 0.033 MoM identified patients who delivered within 2 weeks because of PE with a sensitivity of 93% (25/27) and a specificity of 78% (25/32). This cut-off was associated with a shorter interval-to-delivery due to PE [hazard ratio = 6 (95% CI 2.5–14.6)]. Conclusions: Plasma concentrations of angiogenic/anti-angiogenic factors are of prognostic value in the obstetrical triage area. These observations support the value of these biomarkers in the clinical setting for the identification of the patient at risk for disease progression requiring preterm delivery.

Onto-Tools: new additions and improvements in 2006

Onto-Tools is a freely available web-accessible software suite, composed of an annotation database and nine complementary data-mining tools. This article describes a new tool, Onto-Express-to-go (OE2GO), as well as some new features implemented in Pathway-Express and Onto-Miner over the past year. Pathway-Express (PE) has been enhanced to identify significantly perturbed pathways in a given condition using the differentially expressed genes in the input. OE2GO is a tool for functional profiling using custom annotations. The development of this tool was aimed at the researchers working with organisms for which annotations are not yet available in the public domain. OE2GO allows researchers to use either annotation data from the Onto-Tools database, or their own custom annotations. By removing the necessity to use any specific database, OE2GO makes the functional profiling available for all organisms, with annotations using any ontology. The Onto-Tools are freely available at http://vortex.cs.wayne.edu/projects.htm.

The Transcriptome of the Fetal Inflammatory Response Syndrome

Problem  The fetal inflammatory response syndrome (FIRS) is considered the counterpart of the systemic inflammatory response syndrome (SIRS), but similarities in their regulatory mechanisms are unclear. This study characterizes the fetal mRNA transcriptome of peripheral leukocytes to identify key biological processes and pathways involved in FIRS.

Placental protein 13 (galectin-13) has decreased placental expression but increased shedding and maternal serum concentrations in patients presenting with preterm pre-eclampsia and HELLP syndrome

Placental protein 13 (PP13) is a galectin expressed by the syncytiotrophoblast. Women who subsequently develop preterm pre-eclampsia have low first trimester maternal serum PP13 concentrations. This study revealed that third trimester maternal serum PP13 concentration increased with gestational age in normal pregnancies (p < 0.0001), and it was significantly higher in women presenting with preterm pre-eclampsia (p = 0.02) and hemolysis, elevated liver enzymes, and low platelet count (HELLP) syndrome (p = 0.01) than in preterm controls. Conversely, placental PP13 mRNA (p = 0.03) and protein, as well as cytoplasmic PP13 staining of the syncytiotrophoblast (p < 0.05) was decreased in these pathological pregnancies compared to controls. No differences in placental expression and serum concentrations of PP13 were found at term between patients with pre-eclampsia and control women. In contrast, the immunoreactivity of the syncytiotrophoblast microvillous membrane was stronger in both term and preterm pre-eclampsia and HELLP syndrome than in controls. Moreover, large syncytial cytoplasm protrusions, membrane blebs and shed microparticles strongly stained for PP13 in pre-eclampsia and HELLP syndrome. In conclusion, parallel to its decreased placental expression, an augmented membrane shedding of PP13 contributes to the increased third trimester maternal serum PP13 concentrations in women with preterm pre-eclampsia and HELLP syndrome.

Autoantibody Approach for Serum-Based Detection of Head and Neck Cancer

Currently, no effective tool exists for screening or early diagnosis of head and neck squamous cell carcinoma (HNSCC). Here, we describe an approach for cancer detection based on analysis of patterns of serum immunoreactivity against a panel of biomarkers selected using microarray-based serologic profiling and specialized bioinformatics. We biopanned phage display libraries derived from three different HNSCC tissues to generate 5,133 selectively cloned tumor antigens. Based on their differential immunoreactivity on protein microarrays against serum immunoglobulins from 39 cancer and 41 control patients, we reduced the number of clones to 1,021. The performance of a neural network model (Multilayer Perceptron) for cancer classification on a data set of 80 HNSCC and 78 control samples was assessed using 10-fold cross-validation repeated 100 times. A panel of 130 clones was found to be adequate for building a classifier with sufficient sensitivity and specificity. Using these 130 markers on a completely new and independent set of 80 samples, an accuracy of 84.9% with sensitivity of 79.8% and specificity of 90.1% was achieved. Similar performance was achieved by reshuffling of the data set and by using other classification models. The performance of this classification approach represents a significant improvement over current diagnostic accuracy (sensitivity of 37% to 46% and specificity of 24%) in the primary care setting. The results shown here are promising and show the potential use of this approach toward eventual development of diagnostic assay with sufficient sensitivity and specificity suitable for detection of early-stage HNSCC in high-risk populations. (Cancer Epidemiol Biomarkers Prev 2007;16(11):2396–405)