Federico M. Stefanini

Combining multi-dimensional scaling and cluster analysis to describe the diversity of rural households

Capturing agricultural heterogeneity through the analysis of farm typologies is key with regard to the design of sustainable policies and to the adoptability of new technologies. An optimal balance needs to be found between, on the one hand, the requirement to consider local stakeholder and expert knowledge for typology identification, and on the other hand, the need to identify typologies that transcend the local boundaries of single studies and can be used for comparisons. In this paper, we propose a method that supports expert-driven identification of farm typologies, while at the same time keeping the characteristics of objectivity and reproducibility of statistical tools. The method uses a range of multivariate analysis techniques and it is based on a protocol that favours the use of stakeholder and expert knowledge in the process of typology identification by means of visualization of farm groups and relevant statistics. Results of two studies in Zimbabwe and Kenya are shown. Findings obtained with the method proposed are contrasted with those obtained through a parametric method based on latent class analysis. The method is compared to alternative approaches with regard to stakeholder-orientation and statistical reliability.

Image analysis and automatic classification of transformed foci

Carcinogenesis is a multi‐step process involving genetic alterations and non‐genotoxic mechanisms. The in vitro cell transformation assay allows the monitoring of the neoplastic phenotype by foci formation in suitable cells (e.g. C3H10T1/2 mouse embryo fibroblasts) showing aberrant morphology of massive build‐up, polar and multi‐layered densely stained cells. The classification of transformed foci in C3H cells relies on light microscopy scoring by a trained human expert based on standard rules. This procedure is time‐consuming and prone, in some cases, to subjectivity, thereby leading to possible over‐ or under‐estimation of the carcinogenic potential of tested compounds. Herewith we describe the in vitro neoplastic transformation induced by B[a]P and CdCl2, and the development of a foci classifier based on image analysis and statistical classification. The image analysis system, which relies on ‘spectrum enhancement’, is quantitative and extracts descriptors of foci texture and structure. The statistical classification method is based on the Random Forest algorithm. We obtained a classifier trained by using experts supervision with a 20% classification error. The proposed method could serve as a basis to automate the in vitro cell transformation assay.

The in vitro effect of gossypol and its interaction with salts on conidial germination and viability of Fusarium oxysporum sp. vasinfectum isolates

Aims: To assess the effect of different concentrations of gossypol (0, 2, 4, 10 and 20 mg l–1) in combination with NaCl and Na2SO4 (20 mS cm–1) on the conidial germination and viability of Fusarium oxysporum f.sp. vasinfectum (Fov).

Variability in the response toPseudomonas solanacearum of transgenic lines of potato carrying a cecropin gene analogue

SummaryTransgenic potato plants of cv. Désirée carrying an antibacterial gene, coding for a cecropin lytic peptide analogue, were inoculated with a virulent strain ofPseudomonas solanacearum under controlled conditions. The disease index scored during three repeated infection trials indicated an increased variability in plant response among the transgenic lines which gave either a more susceptible or a more resistant response to the pathogen when compared with untransformed Désirée. Immunity toP. solanacearum was not observed, but it was possible to select a group of transgenic lines that showed resistance levels and disease development curves comparable to the field resistant cv. Cruza 148.

The reduction of large molecular profiles to informative components using a Genetic Algorithm

MOTIVATION Molecular profiles (DNA fingerprints) may be used to allocate an individual of unknown membership to one among the known groups of a reference population. Time and costs of profile assessment may be reduced by identifying informative profile components (markers). RESULTS A genetic algorithm (GA) is proposed to identify promising candidate markers from a pilot experiment in which observations are supposed to be without measurement error. The analysis of simulated datasets suggests reasonable values for GA parameters and confirms that the GA finds components of the profile showing association with the considered groups. Our GA may be used to perform a first screening of candidate markers to be included in subsequent experiments. AVAILABILITY The 32-bit executable (Windows 95, 98 and NT) is available at http://www.ds.unifi.it/ approximately stefanin/bioinformatics.htm.

Objective scoring of transformed foci in BALB/c 3T3 cell transformation assay by statistical image descriptors

In vitro cell transformation assays (CTAs) have been shown to model important stages of in vivo carcinogenesis and have the potential to predict carcinogenicity in humans. Advantages of CTAs are their ability of revealing both genotoxic and non-genotoxic carcinogens while reducing both experimental costs and the number of animals used. The endpoint of the CTA is foci formation, and requires classification under light microscopy based on morphology. Thus current limitations for the wide adoption of the assay partially depend on a fair degree of subjectivity in foci scoring. An objective evaluation may be obtained after separating foci from background monolayer in the digital image, and quantifying values of statistical descriptors which are selected to capture eye-scored morphological features. The aim of this study was to develop statistical descriptors to be applied to transformed foci of BALB/c 3T3, which cover foci size, multilayering and invasive cell growth into the background monolayer. Proposed descriptors were applied to a database of 407 foci images to explore the numerical features, and to illustrate open problems and potential solutions.

Evaluation of resistance to Pseudomonas solanacearum in potato under controlled conditions

SummaryPotato plantlets derived from in vitro propagation of three cultivars known for their field resistance (Cruza 148 and BR-63.65) or susceptibility (Désirée) to Pseudomonas solanacearum E.F. Smith were artificially inoculated under controlled conditions. The aim of this work was to determine the optimal inoculum concentration and the best observation period in which the cultivars would show different reactions to bacterial infection as expected on the basis of their field performance.A suitable statistical analysis of disease indices is proposed to distinguish between resistant and susceptible responses, with a particular care for the applicative needs and a univocal interpretation of the results. In order to evaluate the significance of sources of variation related to the observed mean differences, the analysis of variance and a convenient clustering procedure of disease index means were applied.The statistical analysis revealed that, under our conditions, an inoculum concentration of 5×106 cfu/plant was suitable for separating resistant from susceptible responses, in accordance with the reactions already observed in field experiments by other authors. Also, differences among the three cultivars were best observed nine to twelve days after inoculation with the pathogen.

Using weak prior information on structures to learn Bayesian networks

Most of the approaches developed in the literature to elicit the apriori distribution on Directed Acyclic Graphs (DAGs) require a full specification of graphs. Nevertheless, experts prior knowledge about conditional independence relations may be weak, making the elicitation task troublesome. Moreover, the detailed specification of prior distributions for structural learning is NP-Hard, making the elicitation of large networks impractical. This is the case, for example, of gene expression analysis, in which a small degree of graph connectivity is a priori plausible and where substantial information may regard dozens against thousands of nodes. In this paper we propose an elicitation procedure for DAGs which exploits prior knowledge on network topology, and that is suited to large Bayesian Networks. Then, we develop a new quasi-Bayesian score function, the P-metric, to perform structural learning following a score-and-search approach.

M-GA: A Genetic Algorithm to Search for the Best Conditional Gaussian Bayesian Network

The search of optimal Bayesian network from a database of observations is NP-hard. Nevertheless, several heuristic search strategies have been found to be effective. We present a new population-based algorithm to learn the structure of Bayesian networks without assuming any ordering of nodes and allowing for the presence of both discrete and continuous random variables. Numerical performances of our mixed-genetic algorithm, (M-GA), are investigated on a case study taken from the literature and compared with greedy search

Cadmium-transformed cells in the in vitro cell transformation assay reveal different proliferative behaviours and activated pathways.

The in vitro Cell Transformation Assay (CTA) is a powerful tool for mechanistic studies of carcinogenesis. The endpoint is the classification of transformed colonies (foci) by means of standard morphological features. To increase throughput and reliability of CTAs, one of the suggested follow-up activities is to exploit the comprehension of the mechanisms underlying cell transformation. To this end, we have performed CTAs testing CdCl2, a widespread environmental contaminant classified as a human carcinogen with the underlying mechanisms of action not completely understood. We have isolated and re-seeded the cells at the end (6weeks) of in vitro CTAs to further identify the biochemical pathways underlying the transformed phenotype of foci. Morphological evaluations and proliferative assays confirmed the loss of contact-inhibition and the higher proliferative rate of transformed clones. The biochemical analysis of EGFR pathway revealed that, despite the same initial carcinogenic stimulus (1μM CdCl2 for 24h), transformed clones are characterized by the activation of two different molecular pathways: proliferation (Erk activation) or survival (Akt activation). Our preliminary results on molecular characterization of cell clones from different foci could be exploited for CTAs improvement, supporting the comprehension of the in vivo process and complementing the morphological evaluation of foci.