Machine learning algorithms for spatial data. Case studies: Environmental pollution, natural hazards, renewable resources

This paper presents general problems and approaches for the spatial data analysis using machine learning algorithms. Machine learning is a very powerful approach to adaptive data analysis, modelling and visualisation. The key feature of the machine learning algorithms is that they learn from empirical data and can be used in cases when the modelled environmental phenomena are hidden, nonlinear, noisy and highly variable in space and in time. Most of the machines learning algorithms are universal and adaptive modelling tools developed to solve basic problems of learning from data: classification/pattern recognition, regression/mapping and probability density modelling. In the present report some of the widely used machine learning algorithms, namely artificial neural networks (ANN) of different architectures and Support Vector Machines (SVM), are adapted to the problems of the analysis and modelling of geo-spatial data. Machine learning algorithms have an important advantage over traditional models of spatial statistics when problems are considered in a high dimensional geo-feature spaces, when the dimension of space exceeds 5. Such features are usually generated, for example, from digital elevation models, remote sensing images, etc. An important extension of models concerns considering of real space constrains like geomorphology, networks, and other natural structures. Recent developments in semi-supervised learning can improve modelling of environmental phenomena taking into account on geo-manifolds. An important part of the study deals with the analysis of relevant variables and models' inputs. This problem is approached by using different feature selection/feature extraction nonlinear tools. To demonstrate the application of machine learning algorithms several interesting case studies are considered: digital soil mapping using SVM, automatic mapping of soil and water system pollution using ANN; natural hazards risk analysis (avalanches, landslides), assessments of renewable resources (wind fields) with SVM and ANN models, etc. The dimensionality of spaces considered varies from 2 to more than 30. Figures 1, 2, 3 demonstrate some results of the studies and their outputs. Finally, the results of environmental mapping are discussed and compared with traditional models of geostatistics.

Exogenous selection shapes germination behaviour and seedling traits of populations at different altitudes in a Senecio hybrid zone.

BACKGROUND AND AIMS: The Senecio hybrid zone on Mt Etna, Sicily, is characterized by steep altitudinal clines in quantitative traits and genetic variation. Such clines are thought to be maintained by a combination of 'endogenous' selection arising from genetic incompatibilities and environment-dependent 'exogenous' selection leading to local adaptation. Here, the hypothesis was tested that local adaptation to the altitudinal temperature gradient contributes to maintaining divergence between the parental species, S. chrysanthemifolius and S. aethnensis. METHODS: Intra- and inter-population crosses were performed between five populations from across the hybrid zone and the germination and early seedling growth of the progeny were assessed. KEY RESULTS: Seedlings from higher-altitude populations germinated better under low temperatures (9-13 °C) than those from lower altitude populations. Seedlings from higher-altitude populations had lower survival rates under warm conditions (25/15 °C) than those from lower altitude populations, but also attained greater biomass. There was no altitudinal variation in growth or survival under cold conditions (15/5 °C). Population-level plasticity increased with altitude. Germination, growth and survival of natural hybrids and experimentally generated F(1)s generally exceeded the worse-performing parent. CONCLUSIONS: Limited evidence was found for endogenous selection against hybrids but relatively clear evidence was found for divergence in seed and seedling traits, which is probably adaptive. The combination of low-temperature germination and faster growth in warm conditions might enable high-altitude S. aethnensis to maximize its growth during a shorter growing season, while the slower growth of S. chrysanthemifolius may be an adaptation to drought stress at low altitudes. This study indicates that temperature gradients are likely to be an important environmental factor generating and maintaining adaptive divergence across the Senecio hybrid zone on Mt Etna.

Highly diverse TCRα chain repertoire of pre-immune CD8(+) T cells reveals new insights in gene recombination.

Although the T-cell receptor αδ (TCRαδ) locus harbours large libraries of variable (TRAV) and junctional (TRAJ) gene segments, according to previous studies the TCRα chain repertoire is of limited diversity due to restrictions imposed by sequential coordinate TRAV-TRAJ recombinations. By sequencing tens of millions of TCRα chain transcripts from naive mouse CD8(+) T cells, we observed a hugely diverse repertoire, comprising nearly all possible TRAV-TRAJ combinations. Our findings are not compatible with sequential coordinate gene recombination, but rather with a model in which contraction and DNA looping in the TCRαδ locus provide equal access to TRAV and TRAJ gene segments, similarly to that demonstrated for IgH gene recombination. Generation of the observed highly diverse TCRα chain repertoire necessitates deletion of failed attempts by thymic-positive selection and is essential for the formation of highly diverse TCRαβ repertoires, capable of providing good protective immunity.

Missing value imputation in longitudinal measures of alcohol consumption

Attrition in longitudinal studies can lead to biased results. The study is motivated by the unexpected observation that alcohol consumption decreased despite increased availability, which may be due to sample attrition of heavy drinkers. Several imputation methods have been proposed, but rarely compared in longitudinal studies of alcohol consumption. The imputation of consumption level measurements is computationally particularly challenging due to alcohol consumption being a semi-continuous variable (dichotomous drinking status and continuous volume among drinkers), and the non-normality of data in the continuous part. Data come from a longitudinal study in Denmark with four waves (2003-2006) and 1771 individuals at baseline. Five techniques for missing data are compared: Last value carried forward (LVCF) was used as a single, and Hotdeck, Heckman modelling, multivariate imputation by chained equations (MICE), and a Bayesian approach as multiple imputation methods. Predictive mean matching was used to account for non-normality, where instead of imputing regression estimates, "real" observed values from similar cases are imputed. Methods were also compared by means of a simulated dataset. The simulation showed that the Bayesian approach yielded the most unbiased estimates for imputation. The finding of no increase in consumption levels despite a higher availability remained unaltered. Copyright (C) 2011 John Wiley & Sons, Ltd.

Mechanisms governing lentivirus integration site selection.

An essential step of the life cycle of retroviruses is the stable insertion of a copy of their DNA genome into the host cell genome, and lentiviruses are no exception. This integration step, catalyzed by the viral-encoded integrase, ensures long-term expression of the viral genes, thus allowing a productive viral replication and rendering retroviral vectors also attractive for the field of gene therapy. At the same time, this ability to integrate into the host genome raises safety concerns regarding the use of retroviral-based gene therapy vectors, due to the genomic locations of integration sites. The availability of the human genome sequence made possible the analysis of the integration site preferences, which revealed to be nonrandom and retrovirus-specific, i.e. all lentiviruses studied so far favor integration in active transcription units, while other retroviruses have a different integration site distribution. Several mechanisms have been proposed that may influence integration targeting, which include (i) chromatin accessibility, (ii) cell cycle effects, and (iii) tethering proteins. Recent data provide evidence that integration site selection can occur via a tethering mechanism, through the recruitment of the lentiviral integrase by the cellular LEDGF/p75 protein, both proteins being the two major players in lentiviral integration targeting.

Assessing the Relationship between the Baseline Value of a Continuous Variable and Subsequent Change over Time

Analyzing the relationship between the baseline value and subsequent change of a continuous variable is a frequent matter of inquiry in cohort studies. These analyses are surprisingly complex, particularly if only two waves of data are available. It is unclear for non-biostatisticians where the complexity of this analysis lies and which statistical method is adequate.With the help of simulated longitudinal data of body mass index in children,we review statistical methods for the analysis of the association between the baseline value and subsequent change, assuming linear growth with time. Key issues in such analyses are mathematical coupling, measurement error, variability of change between individuals, and regression to the mean. Ideally, it is better to rely on multiple repeated measurements at different times and a linear random effects model is a standard approach if more than two waves of data are available. If only two waves of data are available, our simulations show that Blomqvist's method - which consists in adjusting for measurement error variance the estimated regression coefficient of observed change on baseline value - provides accurate estimates. The adequacy of the methods to assess the relationship between the baseline value and subsequent change depends on the number of data waves, the availability of information on measurement error, and the variability of change between individuals.

Liver Transplantation for Neuroendocrine Tumors in Europe-Results and Trends in Patient Selection: A 213-Case European Liver Transplant Registry Study.

OBJECTIVE:: The purpose of this study was to assess outcomes and indications in a large cohort of patients who underwent liver transplantation (LT) for liver metastases (LM) from neuroendocrine tumors (NET) over a 27-year period. BACKGROUND:: LT for NET remains controversial due to the absence of clear selection criteria and the scarcity and heterogeneity of reported cases. METHODS:: This retrospective multicentric study included 213 patients who underwent LT for NET performed in 35 centers in 11 European countries between 1982 and 2009. One hundred seven patients underwent transplantation before 2000 and 106 after 2000. Mean age at the time of LT was 46 years. Half of the patients presented hormone secretion and 55% had hepatomegaly. Before LT, 83% of patients had undergone surgical treatment of the primary tumor and/or LM and 76% had received chemotherapy. The median interval between diagnosis of LM and LT was 25 months (range, 1-149 months). In addition to LT, 24 patients underwent major resection procedures and 30 patients underwent minor resection procedures. RESULTS:: Three-month postoperative mortality was 10%. At 5 years after LT, overall survival (OS) was 52% and disease-free survival was 30%. At 5 years from diagnosis of LM, OS was 73%. Multivariate analysis identified 3 predictors of poor outcome, that is, major resection in addition to LT, poor tumor differentiation, and hepatomegaly. Since 2000, 5-year OS has increased to 59% in relation with fewer patients presenting poor prognostic factors. Multivariate analysis of the 106 cases treated since 2000 identified the following predictors of poor outcome: hepatomegaly, age more than 45 years, and any amount of resection concurrent with LT. CONCLUSIONS:: LT is an effective treatment of unresectable LM from NET. Patient selection based on the aforementioned predictors can achieve a 5-year OS between 60% and 80%. However, use of overly restrictive criteria may deny LT to some patients who could benefit. Optimal timing for LT in patients with stable versus progressive disease remains unclear.

Indicators of inequality: Classification and selection of ethnic health disparity indicators

Classification and selection of ethnic disparity health indicators in New Zealand

Vascular-endothelial-growth-factor (VEGF) targeting therapies for endocrine refractory or resistant metastatic breast cancer.

BACKGROUND: Vascular-endothelial-growth-factor (VEGF) is a key mediator of angiogenesis. VEGF-targeting therapies have shown significant benefits and been successfully integrated in routine clinical practice for other types of cancer, such as metastatic colorectal cancer. By contrast, individual trial results in metastatic breast cancer (MBC) are highly variable and their value is controversial. OBJECTIVES: To evaluate the benefits (in progression-free survival (PFS) and overall survival (OS)) and harms (toxicity) of VEGF-targeting therapies in patients with hormone-refractory or hormone-receptor negative metastatic breast cancer. SEARCH METHODS: Searches of CENTRAL, MEDLINE, EMBASE, the Cochrane Breast Cancer Group's Specialised Register, registers of ongoing trials and proceedings of conferences were conducted in January and September 2011, starting in 2000. Reference lists were scanned and members of the Cochrane Breast Cancer Group, experts and manufacturers of relevant drug were contacted to obtain further information. No language restrictions were applied. SELECTION CRITERIA: Randomised controlled trials (RCTs) to evaluate treatment benefit and non-randomised studies in the routine oncology practice setting to evaluate treatment harms. DATA COLLECTION AND ANALYSIS: We performed data collection and analysis according to the published protocol. Individual patient data was sought but not provided. Therefore, the meta-analysis had to be based on published data. Summary statistics for the primary endpoint (PFS) were hazard ratios (HRs). MAIN RESULTS: We identified seven RCTs, one register, and five ongoing trials from a total of 347 references. The published trials for VEGF-targeting drugs in MBC were limited to bevacizumab. Four trials, including a total of 2886 patients, were available for the comparison of first-line chemotherapy, with versus without bevacizumab. PFS (HR 0.67; 95% confidence interval (CI) 0.61 to 0.73) and response rate were significantly better for patients treated with bevacizumab, with moderate heterogeneity regarding the magnitude of the effect on PFS. For second-line chemotherapy, a smaller, but still significant benefit in terms of PFS could be demonstrated for patients treated with bevacizumab (HR 0.85; 95% CI 0.73 to 0.98), as well as a benefit in tumour response. However, OS did not differ significantly, neither in first- (HR 0.93; 95% CI 0.84 to 1.04), nor second-line therapy (HR 0.98; 95% CI 0.83 to 1.16). Quality of life (QoL) was evaluated in four trials but results were published for only two of these with no relevant impact. Subgroup analysis stated a significant greater benefit for patients with previous (taxane) chemotherapy and patients with hormone-receptor negative status. Regarding toxicity, data from RCTs and registry data were consistent and in line with the known toxicity profile of bevacizumab. While significantly higher rates of adverse events (AEs) grade III/IV (odds ratio (OR) 1.77; 95% CI 1.44 to 2.18) and serious adverse events (SAEs) (OR 1.41; 95% CI 1.13 to 1.75) were observed in patients treated with bevacizumab, rates of treatment-related deaths were lower in patients treated with bevacizumab (OR 0.60; 95% CI 0.36 to 0.99). AUTHORS' CONCLUSIONS: The overall patient benefit from adding bevacizumab to first- and second-line chemotherapy in metastatic breast cancer can at best be considered as modest. It is dependent on the type of chemotherapy used and limited to a prolongation of PFS and response rates in both first- and second-line therapy, both surrogate parameters. In contrast, bevacizumab has no significant impact on the patient-related secondary outcomes of OS or QoL, which indicate a direct patient benefit. For this reason, the clinical value of bevacizumab for metastatic breast cancer remains controversial.

Vaccination with a Melan-A peptide selects an oligoclonal T cell population with increased functional avidity and tumor reactivity.

Both the underlying molecular mechanisms and the kinetics of TCR repertoire selection following vaccination against tumor Ags in humans have remained largely unexplored. To gain insight into these questions, we performed a functional and structural longitudinal analysis of the TCR of circulating CD8(+) T cells specific for the HLA-A2-restricted immunodominant epitope from the melanocyte differentiation Ag Melan-A in a melanoma patient who developed a vigorous and sustained Ag-specific T cell response following vaccination with the corresponding synthetic peptide. We observed an increase in functional avidity of Ag recognition and in tumor reactivity in the postimmune Melan-A-specific populations as compared with the preimmune blood sample. Improved Ag recognition correlated with an increase in the t(1/2) of peptide/MHC interaction with the TCR as assessed by kinetic analysis of A2/Melan-A peptide multimer staining decay. Ex vivo analysis of the clonal composition of Melan-A-specific CD8(+) T cells at different time points during vaccination revealed that the response was the result of asynchronous expansion of several distinct T cell clones. Some of these T cell clones were also identified at a metastatic tumor site. Collectively, these data show that tumor peptide-driven immune stimulation leads to the selection of high-avidity T cell clones of increased tumor reactivity that independently evolve within oligoclonal populations.

Clonotype selection and composition of human CD8 T cells specific for persistent herpes viruses varies with differentiation but is stable over time.

Protection from reactivation of persistent herpes virus infection is mediated by Ag-specific CD8 T cell responses, which are highly regulated by still poorly understood mechanisms. In this study, we analyzed differentiation and clonotypic dynamics of EBV- and CMV-specific T cells from healthy adults. Although these T lymphocytes included all subsets, from early-differentiated (EM/CD28(pos)) to late-differentiated (EMRA/CD28(neg)) stages, they varied in the sizes/proportions of these subsets. In-depth clonal composition analyses revealed TCR repertoires, which were highly restricted for CMV- and relatively diverse for EBV-specific cells. Virtually all virus-specific clonotypes identified in the EMRA/CD28(neg) subset were also found within the pool of less differentiated "memory" cells. However, striking differences in the patterns of dominance were observed among these subsets, because some clonotypes were selected with differentiation while others were not. Late-differentiated CMV-specific clonotypes were mostly characterized by TCR with lower dependency on CD8 coreceptor interaction. Yet all clonotypes displayed similar functional avidities, suggesting a compensatory role of CD8 in the clonotypes of lower TCR avidity. Importantly, clonotype selection and composition of each virus-specific subset upon differentiation was highly preserved over time, with the presence of the same dominant clonotypes at specific differentiation stages within a period of 4 years. Remarkably, clonotypic distribution was stable not only in late-differentiated but also in less-differentiated T cell subsets. Thus, T cell clonotypes segregate with differentiation, but the clonal composition once established is kept constant for at least several years. These findings reveal novel features of the highly sophisticated control of steady state protective T cell activity in healthy adults.

The robustness of the weak selection approximation for the evolution of altruism against strong selection.

The weak selection approximation of population genetics has made possible the analysis of social evolution under a considerable variety of biological scenarios. Despite its extensive usage, the accuracy of weak selection in predicting the emergence of altruism under limited dispersal when selection intensity increases remains unclear. Here, we derive the condition for the spread of an altruistic mutant in the infinite island model of dispersal under a Moran reproductive process and arbitrary strength of selection. The simplicity of the model allows us to compare weak and strong selection regimes analytically. Our results demonstrate that the weak selection approximation is robust to moderate increases in selection intensity and therefore provides a good approximation to understand the invasion of altruism in spatially structured population. In particular, we find that the weak selection approximation is excellent even if selection is very strong, when either migration is much stronger than selection or when patches are large. Importantly, we emphasize that the weak selection approximation provides the ideal condition for the invasion of altruism, and increasing selection intensity will impede the emergence of altruism. We discuss that this should also hold for more complicated life cycles and for culturally transmitted altruism. Using the weak selection approximation is therefore unlikely to miss out on any demographic scenario that lead to the evolution of altruism under limited dispersal.

Targeted expression of human CD1d in transgenic mice reveals independent roles for thymocytes and thymic APCs in positive and negative selection of Valpha14i NKT cells.

CD1d-dependent invariant Valpha14 (Valpha14i) NKT cells are innate T lymphocytes expressing a conserved semi-invariant TCR, consisting, in mice, of the invariant Valpha14-Jalpha18 TCR alpha-chain paired mostly with Vbeta8.2 and Vbeta7. The cellular requirements for thymic positive and negative selection of Valpha14i NKT cells are only partially understood. Therefore, we generated transgenic mice expressing human CD1d (hCD1d) either on thymocytes, mainly CD4+ CD8+ double positive, or on APCs, the cells implicated in the selection of Valpha14i NKT cells. In the absence of the endogenous mouse CD1d (mCD1d), the expression of hCD1d on thymocytes, but not on APCs, was sufficient to select Valpha14i NKT cells that proved functional when activated ex vivo with the Ag alpha-galactosyl ceramide. Valpha14i NKT cells selected by hCD1d on thymocytes, however, attained lower numbers than in control mice and expressed essentially Vbeta8.2. The low number of Vbeta8.2+ Valpha14i NKT cells selected by hCD1d on thymocytes was not reversed by the concomitant expression of mCD1d, which, instead, restored the development of Vbeta7+ Valpha14i NKT cells. Vbeta8.2+, but not Vbeta7+, NKT cell development was impaired in mice expressing both hCD1d on APCs and mCD1d. Taken together, our data reveal that selective CD1d expression by thymocytes is sufficient for positive selection of functional Valpha14i NKT cells and that both thymocytes and APCs may independently mediate negative selection.

Selection of peptides and synthesis of pentameric peptabody molecules reacting specifically with ErbB-2 receptor.

The HER-2/ErbB-2 oncoprotein is overexpressed in human breast and ovarian adenocarcinomas and is clearly associated with the malignant phenotype. Although no specific ligand for this receptor has been positively identified, ErbB-2 was shown to play a central role in a network of interactions with the related ErbB-1, ErbB-3 and ErbB-4 receptors. We have selected new peptides binding to ErbB-2 extracellular domain protein (ECD) by screening 2 newly developed constrained and unconstrained random hexapeptide phage libraries. Out of 37 phage clones, which bound specifically to ErbB-2 ECD, we found 6 constrained and 10 linear different hexapeptide sequences. Among the latter, 5 consensus motifs, all with a common methionine and a positively charged residue at positions 1 and 3, respectively, were identified. Furthermore, 3 representative hexapeptides were fused to a coiled-coil pentameric recombinant protein to form the so-called peptabodies recently developed in our laboratory. The 3 peptabodies bound specifically to the ErbB-2 ECD, as determined by enzyme-linked immunosorbent assay and BIAcore analysis and to tumor cells overexpressing ErbB-2, as shown by flow cytometry. Interestingly, one of the free selected linear peptides and all 3 peptabodies inhibited the proliferation of tumor cells overexpressing ErbB-2. In conclusion, a novel type of ErbB-2-specific ligand is described that might complement presently available monoclonal antibodies.