Cluster kernels for semisupervised classification of VHR urban images

In this paper, we present and apply a semisupervised support vector machine based on cluster kernels for the problem of very high resolution image classification. In the proposed setting, a base kernel working with labeled samples only is deformed by a likelihood kernel encoding similarities between unlabeled examples. The resulting kernel is used to train a standard support vector machine (SVM) classifier. Experiments carried out on very high resolution (VHR) multispectral and hyperspectral images using very few labeled examples show the relevancy of the method in the context of urban image classification. Its simplicity and the small number of parameters involved make it versatile and workable by unexperimented users.

The arbuscular mycorrhizal fungus Glomus intraradices is haploid and has a small genome size in the lower limit of eukaryotes.

The genome size, complexity, and ploidy of the arbuscular mycorrhizal fungus (AMF) Glomus intraradices was determined using flow cytometry, reassociation kinetics, and genomic reconstruction. Nuclei of G. intraradices from in vitro culture, were analyzed by flow cytometry. The estimated average length of DNA per nucleus was 14.07+/-3.52 Mb. Reassociation kinetics on G. intraradices DNA indicated a haploid genome size of approximately 16.54 Mb, comprising 88.36% single copy DNA, 1.59% repetitive DNA, and 10.05% fold-back DNA. To determine ploidy, the DNA content per nucleus measured by flow cytometry was compared with the genome estimate of reassociation kinetics. G. intraradices was found to have a DNA index (DNA per nucleus per haploid genome size) of approximately 0.9, indicating that it is haploid. Genomic DNA of G. intraradices was also analyzed by genomic reconstruction using four genes (Malate synthase, RecA, Rad32, and Hsp88). Because we used flow cytometry and reassociation kinetics to reveal the genome size of G. intraradices and show that it is haploid, then a similar value for genome size should be found when using genomic reconstruction as long as the genes studied are single copy. The average genome size estimate was 15.74+/-1.69 Mb indicating that these four genes are single copy per haploid genome and per nucleus of G. intraradices. Our results show that the genome size of G. intraradices is much smaller than estimates of other AMF and that the unusually high within-spore genetic variation that is seen in this fungus cannot be due to high ploidy.

The demographic benefits of belligerence and bravery: defeated group repopulation or victorious group size expansion?

Intraspecific coalitional aggression between groups of individuals is a widespread trait in the animal world. It occurs in invertebrates and vertebrates, and is prevalent in humans. What are the conditions under which coalitional aggression evolves in natural populations? In this article, I develop a mathematical model delineating conditions where natural selection can favor the coevolution of belligerence and bravery between small-scale societies. Belligerence increases an actor's group probability of trying to conquer another group and bravery increase the actors's group probability of defeating an attacked group. The model takes into account two different types of demographic scenarios that may lead to the coevolution of belligerence and bravery. Under the first, the fitness benefits driving the coevolution of belligerence and bravery come through the repopulation of defeated groups by fission of victorious ones. Under the second demographic scenario, the fitness benefits come through a temporary increase in the local carrying capacity of victorious groups, after transfer of resources from defeated groups to victorious ones. The analysis of the model suggests that the selective pressures on belligerence and bravery are stronger when defeated groups can be repopulated by victorious ones. The analysis also suggests that, depending on the shape of the contest success function, costly bravery can evolve in groups of any size.

Improvement of antibiotic prescription in outpatient care: a cluster-randomized intervention study using a sentinel surveillance network of physicians.

OBJECTIVES: To assess the effectiveness of implementing guidelines, coupled with individual feedback, on antibiotic prescribing behaviour of primary care physicians in Switzerland. METHODS: One hundred and forty general practices from a representative Swiss sentinel network of primary care physicians participated in this cluster-randomized prospective intervention study. The intervention consisted of providing guidelines on treatment of respiratory tract infections (RTIs) and uncomplicated lower urinary tract infections (UTIs), coupled with sustained, regular feedback on individual antibiotic prescription behaviour during 2 years. The main aims were: (i) to increase the percentage of prescriptions of penicillins for all RTIs treated with antibiotics; (ii) to increase the percentage of trimethoprim/sulfamethoxazole prescriptions for all uncomplicated lower UTIs treated with antibiotics; (iii) to decrease the percentage of quinolone prescriptions for all cases of exacerbated COPD (eCOPD) treated with antibiotics; and (iv) to decrease the proportion of sinusitis and other upper RTIs treated with antibiotics. The study was registered at ClinicalTrials.gov (NCT01358916). RESULTS: While the percentage of antibiotics prescribed for sinusitis or other upper RTIs and the percentage of quinolones prescribed for eCOPD did not differ between the intervention group and the control group, there was a significant increase in the percentage of prescriptions of penicillins for all RTIs treated with antibiotics [57% versus 49%, OR=1.42 (95% CI 1.08-1.89), P=0.01] and in the percentage of trimethoprim/sulfamethoxazole prescriptions for all uncomplicated lower UTIs treated with antibiotics [35% versus 19%, OR=2.16 (95% CI 1.19-3.91), P=0.01] in the intervention group. CONCLUSIONS: In our setting, implementing guidelines, coupled with sustained individual feedback, was not able to reduce the proportion of sinusitis and other upper RTIs treated with antibiotics, but increased the use of recommended antibiotics for RTIs and UTIs, as defined by the guidelines.

Assessing the Precision of Compositional Data in a Stratified Double Stage Cluster Sample: Application to the Swiss Earnings Structure Survey

Precision of released figures is not only an important quality feature of official statistics,it is also essential for a good understanding of the data. In this paper we show a casestudy of how precision could be conveyed if the multivariate nature of data has to betaken into account. In the official release of the Swiss earnings structure survey, the totalsalary is broken down into several wage components. We follow Aitchison's approachfor the analysis of compositional data, which is based on logratios of components. Wefirst present diferent multivariate analyses of the compositional data whereby the wagecomponents are broken down by economic activity classes. Then we propose a numberof ways to assess precision

Interactions between grain size and composition of sediments: two examples

Two contrasting case studies of sediment and detrital mineral composition are investigated in order to outline interactions between chemical composition and grain size. Modern glacial sediments exhibit a strong dependence of the two parameters due to the preferential enrichment of mafic minerals, especially biotite, in the fine-grained fractions. On the other hand, the composition of detrital heavy minerals (here: rutile) appears to be not systematically related to grain-size, but is strongly controlled by location, i.e. the petrology of the source rocks of detrital grains. This supports the use of rutile as a well-suited tracer mineral for provenance studies. The results further suggest that (i) interpretations derived from whole-rock sediment geochemistry should be flanked by grain-size observations, and (ii) a more sound statistical evaluation of these interactions require the development of new tailor-made statistical tools to deal with such so-called two-way compositions

Relation of cardiac troponin I measurements at 24 and 48 hours to magnetic resonance-determined infarct size in patients with ST-elevation myocardial infarction.

Levels of circulating cardiac troponin I (cTnI) or T are correlated to extent of myocardial destruction after an acute myocardial infarction. Few studies analyzing this relation have employed a second-generation cTnI assay or cardiac magnetic resonance (CMR) as the imaging end point. In this post hoc study of the Efficacy of FX06 in the Prevention of Mycoardial Reperfusion Injury (F.I.R.E.) trial, we aimed at determining the correlation between single-point cTnI measurements and CMR-estimated infarct size at 5 to 7 days and 4 months after a first-time ST-elevation myocardial infarction (STEMI) and investigating whether cTnI might provide independent prognostic information regarding infarct size at 4 months even taking into account early infarct size. Two hundred twenty-seven patients with a first-time STEMI were included in F.I.R.E. All patients received primary percutaneous coronary intervention within 6 hours from onset of symptoms. cTnI was measured at 24 and 48 hours after admission. CMR was conducted within 1 week of the index event (5 to 7 days) and at 4 months. Pearson correlations (r) for infarct size and cTnI at 24 hours were r = 0.66 (5 days) and r = 0.63 (4 months) and those for cTnI at 48 hours were r = 0.67 (5 days) and r = 0.65 (4 months). In a multiple regression analysis for predicting infarct size at 4 months (n = 141), cTnI and infarct location retained an independent prognostic role even taking into account early infarct size. In conclusion, a single-point cTnI measurement taken early after a first-time STEMI is a useful marker for infarct size and might also supplement early CMR evaluation in prediction of infarct size at 4 months.

Effect of a general school-based physical activity intervention on bone mineral content and density: A cluster-randomized controlled trial.

Background: Specific physical loading leads to enhanced bone development during childhood. A general physical activity program mimicking a real-life situation was successful at increasing general physical health in children. Yet, it is not clear whether it can equally increase bone mineral mass. We performed a cluster-randomized controlled trial in children of both gender and different pubertal stages to determine whether a school-based physical activity (PA) program during one school-year influences bone mineral content (BMC) and density (BMD), irrespective of gender.Methods: Twenty-eight 1st and 5th grade (6-7 and 11-12 year-old) classes were cluster randomized to an intervention (INT, 16 classes, n = 297) and control (CON; 12 classes, n = 205) group. The intervention consisted of a multi-component PA intervention including daily physical education with at least 10 min of jumping or strength training exercises of various intensities. Measurements included anthropometry, and BMC and BMD of total body, femoral neck, total hip and lumbar spine using dual-energy X-ray absorptiometry (DXA). PA was assessed by accelerometers and Tanner stages by questionnaires. Analyses were performed by a regression model adjusted for gender, baseline height and weight, baseline PA, post-intervention pubertal stage, baseline BMC, and cluster.Results: 275 (72%) of 380 children who initially agreed to have DXA measurements had also post-intervention DXA and PA data. Mean age of prepubertal and pubertal children at baseline was 8.7 +/- 2.1 and 11.1 +/- 0.6 years, respectively. Compared to CON, children in INT showed statistically significant increases in BMC of total body, femoral neck, and lumbar spine by 5.5%, 5.4% and 4.7% (all p < 0.05), respectively, and BMD of total body and lumbar spine by 8.4% and 7.3% (both p < 0.01), respectively. There was no gender*group, but a pubertal stage*group interaction consistently favoring prepubertal children.Conclusion: A general school-based PA intervention can increase bone health in elementary school children of both genders, particularly before puberty. (C) 2010 Elsevier Inc. All rights reserved.

A Log-linear model of grain size influence on the geochemistry of sediments

Sediment composition is mainly controlled by the nature of the source rock(s), and chemical (weathering) and physical processes (mechanical crushing, abrasion, hydrodynamic sorting) during alteration and transport. Although the factors controlling these processes are conceptually well understood, detailed quantification of compositional changes induced by a single process are rare, as are examples where the effects of several processes can be distinguished. The present study was designed to characterize the role of mechanical crushing and sorting in the absence of chemical weathering. Twenty sediment samples were taken from Alpine glaciers that erode almost pure granitoid lithologies. For each sample, 11 grain-size fractions from granules to clay (ø grades &-1 to &9) were separated, and each fraction was analysed for its chemical composition.The presence of clear steps in the box-plots of all parts (in adequate ilr and clr scales) against ø is assumed to be explained by typical crystal size ranges for the relevant mineral phases. These scatter plots and the biplot suggest a splitting of the full grain size range into three groups: coarser than ø=4 (comparatively rich in SiO2, Na2O, K2O, Al2O3, and dominated by “felsic” minerals like quartz and feldspar), finer than ø=8 (comparatively rich in TiO2, MnO, MgO, Fe2O3, mostly related to “mafic” sheet silicates like biotite and chlorite), and intermediate grains sizes (4≤ø &8; comparatively rich in P2O5 and CaO, related to apatite, some feldspar).To further test the absence of chemical weathering, the observed compositions were regressed against three explanatory variables: a trend on grain size in ø scale, a step function for ø≥4, and another for ø≥8. The original hypothesis was that the trend could be identified with weathering effects, whereas each step function would highlight those minerals with biggest characteristic size at its lower end. Results suggest that this assumption is reasonable for the step function, but that besides weathering some other factors (different mechanical behavior of minerals) have also an important contribution to the trend.Key words: sediment, geochemistry, grain size, regression, step function

Intra-individual variation in body temperature and pectoral muscle size in nestling Alpine swifts Apus melba in response to an episode of inclement weather

The Alpine swift (Apus melba) forages on insects caught exclusively on the wing, implying that dependent nestlings face acute food shortage in periods of cold and rainy weather. Therefore, there should be strong selection on nestling swifts to evolve physiological strategies to cope with periods of undernutrition. We have investigated intra-individual changes in nestling pectoral muscle and body temperature in response to a 1-week period of inclement weather. The pectoral muscle is the largest reserves of proteins, and nestlings have to devote a large amount of energy in the maintenance of body temperature. The results show that nestling pectoral muscle size and body temperature were significantly reduced during the episode of inclement weather. Assuming that these physiological changes are adaptive, our study suggests that nestling swifts spare energy by a pronounced reduction (up to 18 degrees C) in body temperature and use proteins from the pectoral muscle as a source of extra energy to survive prolonged periods of fasting.

Cluster recognition in spatial-temporal sequences: The case of forest fires

Forest fire sequences can be modelled as a stochastic point process where events are characterized by their spatial locations and occurrence in time. Cluster analysis permits the detection of the space/time pattern distribution of forest fires. These analyses are useful to assist fire-managers in identifying risk areas, implementing preventive measures and conducting strategies for an efficient distribution of the firefighting resources. This paper aims to identify hot spots in forest fire sequences by means of the space-time scan statistics permutation model (STSSP) and a geographical information system (GIS) for data and results visualization. The scan statistical methodology uses a scanning window, which moves across space and time, detecting local excesses of events in specific areas over a certain period of time. Finally, the statistical significance of each cluster is evaluated through Monte Carlo hypothesis testing. The case study is the forest fires registered by the Forest Service in Canton Ticino (Switzerland) from 1969 to 2008. This dataset consists of geo-referenced single events including the location of the ignition points and additional information. The data were aggregated into three sub-periods (considering important preventive legal dispositions) and two main ignition-causes (lightning and anthropogenic causes). Results revealed that forest fire events in Ticino are mainly clustered in the southern region where most of the population is settled. Our analysis uncovered local hot spots arising from extemporaneous arson activities. Results regarding the naturally-caused fires (lightning fires) disclosed two clusters detected in the northern mountainous area.

Comparative genomics of ParaHox clusters of teleost fishes: gene cluster breakup and the retention of gene sets following whole genome duplications.

BACKGROUND: The evolutionary lineage leading to the teleost fish underwent a whole genome duplication termed FSGD or 3R in addition to two prior genome duplications that took place earlier during vertebrate evolution (termed 1R and 2R). Resulting from the FSGD, additional copies of genes are present in fish, compared to tetrapods whose lineage did not experience the 3R genome duplication. Interestingly, we find that ParaHox genes do not differ in number in extant teleost fishes despite their additional genome duplication from the genomic situation in mammals, but they are distributed over twice as many paralogous regions in fish genomes. RESULTS: We determined the DNA sequence of the entire ParaHox C1 paralogon in the East African cichlid fish Astatotilapia burtoni, and compared it to orthologous regions in other vertebrate genomes as well as to the paralogous vertebrate ParaHox D paralogons. Evolutionary relationships among genes from these four chromosomal regions were studied with several phylogenetic algorithms. We provide evidence that the genes of the ParaHox C paralogous cluster are duplicated in teleosts, just as it had been shown previously for the D paralogon genes. Overall, however, synteny and cluster integrity seems to be less conserved in ParaHox gene clusters than in Hox gene clusters. Comparative analyses of non-coding sequences uncovered conserved, possibly co-regulatory elements, which are likely to contain promoter motives of the genes belonging to the ParaHox paralogons. CONCLUSION: There seems to be strong stabilizing selection for gene order as well as gene orientation in the ParaHox C paralogon, since with a few exceptions, only the lengths of the introns and intergenic regions differ between the distantly related species examined. The high degree of evolutionary conservation of this gene cluster's architecture in particular - but possibly clusters of genes more generally - might be linked to the presence of promoter, enhancer or inhibitor motifs that serve to regulate more than just one gene. Therefore, deletions, inversions or relocations of individual genes could destroy the regulation of the clustered genes in this region. The existence of such a regulation network might explain the evolutionary conservation of gene order and orientation over the course of hundreds of millions of years of vertebrate evolution. Another possible explanation for the highly conserved gene order might be the existence of a regulator not located immediately next to its corresponding gene but further away since a relocation or inversion would possibly interrupt this interaction. Different ParaHox clusters were found to have experienced differential gene loss in teleosts. Yet the complete set of these homeobox genes was maintained, albeit distributed over almost twice the number of chromosomes. Selection due to dosage effects and/or stoichiometric disturbance might act more strongly to maintain a modal number of homeobox genes (and possibly transcription factors more generally) per genome, yet permit the accumulation of other (non regulatory) genes associated with these homeobox gene clusters.

Classification of immature mosquito species according to characteristics of the larval habitat in the subtropical province of Chaco, Argentina

To classify mosquito species based on common features of their habitats, samples were obtained fortnightly between June 2001-October 2003 in the subtropical province of Chaco, Argentina. Data on the type of larval habitat, nature of the habitat (artificial or natural), size, depth, location related to sunlight, distance to the neighbouring houses, type of substrate, organic material, vegetation and algae type and their presence were collected. Data on the permanence, temperature, pH, turbidity, colour, odour and movement of the larval habitat's water were also collected. From the cluster analysis, three groups of species associated by their degree of habitat similarity were obtained and are listed below. Group 1 consisted of Aedes aegypti. Group 2 consisted of Culex imitator, Culex davisi, Wyeomyia muehlensi and Toxorhynchites haemorrhoidalis separatus. Within group 3, two subgroups are distinguished: A (Psorophora ferox, Psorophora cyanescens, Psorophora varinervis, Psorophora confinnis, Psorophora cingulata, Ochlerotatus hastatus-oligopistus, Ochlerotatus serratus, Ochlerotatus scapularis, Culex intrincatus, Culex quinquefasciatus, Culex pilosus, Ochlerotatus albifasciatus, Culex bidens) and B (Culex maxi, Culex eduardoi, Culex chidesteri, Uranotaenia lowii, Uranotaenia pulcherrima, Anopheles neomaculipalpus, Anopheles triannulatus, Anopheles albitarsis, Uranotaenia apicalis, Mansonia humeralis and Aedeomyia squamipennis). Principal component analysis indicates that the size of the larval habitats and the presence of aquatic vegetation are the main characteristics that explain the variation among different species. In contrast, water permanence is second in importance. Water temperature, pH and the type of larval habitat are less important in explaining the clustering of species.

Community cluster of meningococcal disease by Neisseria meningitidis serogroup C in Andalusia, Spain, March to May 2011.

Between March and May of 2011, a cluster of three fatal cases of meningococcal sepsis occurred in Andalusia, Spain, in a municipality with a population of around 20,000 inhabitants. The cases were in their mid-teens to early thirties and were notified to the epidemiological surveillance system of Andalusia (Sistema de Vigilancia Epidemiológica de Andalucía, SVEA) during a 68-day period from March through May 2011. All three were infected with the same strain of Neisseria meningitidis serogroup C genosubtype VR1:5-1;VR2:10-8. None of the cases had been previously vaccinated against N. meningitidis serogroup C. Antibiotic post-exposure chemoprophylaxis was administered to close contacts of every diagnosed case. Once the cluster was confirmed, the local population was informed through the media about the control measures taken by the health authorities. The vaccination history against N. meningitidis serogroup C of the population under 25 years-old in the municipality was checked. Vaccination was offered to unimmunised individuals younger than 25 years of age and an additional dose of vaccine was offered to those who had been vaccinated between 2000 and 2006 with a vaccination schedule of three doses before the first year of age. No further cases occurred since the beginning of these actions.

New human kallikrein 14: enzymatic characterization and inhibitors development.

RESUME DESTINE A UN LARGE PUBLIC En biologie, si une découverte permet de répondre à quelques questions, en général elle en engendre beaucoup d'autres. C'est ce qui s'est produit récemment dans le monde des kallicréines. De la famille des protéases, protéines ayant la faculté de couper plus ou moins spécifiquement d'autres protéines pour exercer un rôle biologique, la famille des kallicréines humaines n'était composée que de 3 membres lors du siècle dernier. Parmi eux, une kallicréine mondialement utilisée pour détecter le cancer de la prostate, le PSA. En 2000, un chercheur de l'hôpital universitaire Mont Sinaï à Toronto, le Professeur Eleftherios Diamandis, a découvert la présence de 12 nouveaux gènes appartenant à cette famille, situés sur le même chromosome que les 3 premières kallicréines. Cette découverte majeure a placé les spécialistes des kallicréines face à une montagne d'interrogations car les fonctions de ces nouvelles protéases étaient totalement inconnues. La kallicréine humaine 14 (hK14) présente un intérêt particulier, car elle se retrouve associée à différents cancers, notamment les carcinomes ovariens et mammaires. Cette association ne répond cependant pas à la fonction de cette protéase. L'objectif de ce travail de thèse était donc de découvrir, dans un premier temps, la spécificité de cette nouvelle kallicréine, c'est-à-dire le type de coupure qu'elle engendre au niveau des protéines qu'elle cible. Utilisant une technologie de pointe qui exploite la propriété des bactériophages à se répliquer dans les bactéries à l'infini, des dizaines de millions de combinaisons protéiques aléatoires ont été présentées à hK14, qui a pu sélectionner celles qui lui étaient favorables pour la coupure. Cette technique qualitative porte le nom de Phage Display Substrate. Une fois la sélection réalisée, il fallait transférer ces séquences coupées ou substrats dans un système permettant de donner une valeur quantitative à l'efficacité de coupure. Pour cela nous avons développé une technologie qui permet d'évaluer cette efficacité en utilisant des protéines fluorescentes de méduse, modifiées génétiquement, dont l'excitation de la première (CFP : cyan fluorescent protein) par la lumière à une certaine longue d'onde permet le transfert d'énergie à la seconde (YFP : yellow fluorescent protein), via un substrat qui les lie. Pour que ce transfert d'énergie se produise, il faut que les deux protéines fluorescentes soient proches, comme c'est le cas lorsqu'elles sont liées par un substrat. La coupure de ce lien provoque un changement de transfert d'énergie qui est quantifiable en utilisant un spectrofluoromètre. Cette technologie permet donc de suivre la réaction d'hydrolyse (coupure) des protéases. Afin de poursuivre certaines expériences permettant de mieux comprendre la fonction biologique d'hK14 ainsi que son éventuelle implication dans le cancer, nous avons développé des inhibiteurs spécifiques d'hK14. Les séquences qui on été le plus efficacement coupées par hK14 ont été utilisées pour transformer deux types d'inhibiteurs classiques, qui circulent dans notre sang, en inhibiteurs d'hK14 hautement efficaces et spécifiques. Selon les résultats obtenus in vitro, ils pourront être évalués in vivo en tant que traitement potentiel contre le cancer. RESUME Les protéases sont des enzymes impliquées dans des processus physiologiques mais aussi parfois pathologiques. La famille des kallicréines tissulaires humaines représente le plus grand groupe de protéases humaines, dont plusieurs pourraient participer au développement de certaines maladies. D'autre part, ces protéases sont apparues comme des marqueurs de pathogénicité potentiels, notamment dans les cas de cancers hormono-dépendants. La kallicréine humaine 14 a été récemment découverte et son implication dans quelques maladies, particulièrement dans le cas de tumeurs, semble probable. En effet, son expression génique est augmentée au niveau des tissus cancéreux de la prostate et du sein et son expression protéique s'est révélée plus élevée dans le sérum de patientes atteintes d'un cancer du sein ou des ovaires. Cependant, comme c'est le cas pour la plupart des kallicréines, sa fonction est encore inconnue. Afin de mieux connaître son rôle biologique et/ou pathologique, nous avons décidé de caractériser son activité enzymatique. Nous avons tout d'abord mis au point un système de substrats entièrement biologique permettant d'étudier in vitro l'activité des protéases. Ce système est basé sur le phénomène de FRET, à savoir le transfert d'énergie de résonance fluorescente qui intervient entre deux molécules fluorescentes voisines si le spectre d'émission de la protéine donneuse chevauche le spectre d'excitation de la protéine receveuse. Nous avons fusionné de manière covalente une protéine fluorescente bleue (CFP) et une jaune (YFP) en les liant avec diverses séquences. Par clivage de la séquence de liaison, une perte du transfert d'énergie peut être mesurée par un spectrofluoromètre. Cette technologie représente un moyen facile de suivre la réaction d'hydrolyse des protéases. Les conditions optimales de production de ces substrats CFP-YFP ont été déterminées, de même que les paramètres pouvant éventuellement influencer le FRET. Ce système possède une grande résistance à la protéolyse non spécifique et est applicable à un grand nombre de protéase. Contrairement aux substrats fluorogéniques, il permet d'étudier les acides aminés se trouvant des deux côtés du site de clivage. Ce système étant entièrement biologique, il est le reflet des interactions protéine-protéine et représente un outil biologique facile, bon marché et rapide pour caractériser les protéases. Dans un premier temps, hK14 a été mise en présence d' une banque de haute diversité de pentapeptides aléatoires présentée à la surface de phages afin d'identifier des substrats spécifiques. Ensuite, le système CFP-YFP a été employé pour trier les peptides sélectionnés afin d'identifier les séquences de substrats les plus sensibles et spécifiques pour hK14. Nous avons montré, qu'en plus de sa prévisible activité de type trypsine, hK14 possède aussi une très surprenante activité de type chymotrypsine. Les séquences les plus sensibles ont été choisies pour cribler la banque de donnée Swissprot, permettant ainsi l'identification de 6 substrats protéiques humains potentiels pour hK14. Trois d'entre eux, la laminine α-5, le collagène IV et la matriline-4, qui sont des composants de la matrice extracellulaire, ont démontré une grande susceptibilité à l'hydrolyse par hK14. De plus, la séparation éléctrophorétique a montré que la dégradation de la laminine α-5 et de la matriline-4 par hK14 devait se produire aux sites identifiés par la technologie du phage display. Pour terminer, nous avons transformé, par mutagenèse dirigée, deux serpines (inhibiteurs de protéases de type sérine) connues, AAT et ACT (alpha anti-trypsine et alpha anti-chymotrypsine), qui inhibent un vaste éventail d'enzymes humaines en inhibiteurs d'hK14 hautement efficaces et spécifiques. Ces inhibiteurs pourront être utilisés d'une part pour poursuivre certaines expériences permettant de mieux comprendre l'implication d'hK14 dans des voies physiologiques ou dans le cancer et d'autre part pour les évaluer in vivo en tant que traitement potentiel contre le cancer. SUMMARY Proteases consist of enzymes involved in physiological events, but also, in case of dysregulation, in pathogenicity. The human tissue kallikrein family represents the largest human protease cluster and includes several members that either could participate in the course of certain diseases or emerged as potential biological markers, especially in hormone dependent cancers. The human kallikrein 14 has been recently discovered and suggested implications in some disorders, particularly in tumors since its gene expression is up-regulated in prostate and breast cancer tissues and its protein expression increased in the serum of patients with breast and ovarian cancers. However, like most kallikreins, its function remains unknown. To better understand hK14 biological and/or pathological role, we decided to characterize its enzymatic activity. First of all, we developped a biological system suitable for in vitro study of protease activity. This system is based on the so-called FRET phenomenon, that is the Fluorescence Resonance Energy Transfer that occurs between two nearby fluorescent proteins if the emission spectrum of the donor overlaps the excitation spectrum of the acceptor. We fused covalently a cyan fluorescent protein (CFP) and a yellow fluorescent protein (YFP) with diverses sequences. Upon cleavage of the linker sequence by protease, the loss of energy transfer can be measured by a spectrofluorometer allowing an easy following of hydrolysis reaction. The optimal conditions to produce in bacterial system these CFP-YFP substrates were determined as well as the parameters that could eventually influence the FRET. This system demonstrated a high degree of resistance to non-specific proteolysis and applicability to various conditions corresponding to a great number of existing proteases. Other avantages are the possibility to study the amino acids located both sides of the cleavage site as well as the interest to work in a full biological system reflecting protein-protein interaction. A phage substrate library with exhaustive diversity was used prior to CFP-substrate-YFP system to isolate specific human kallikrein 14 substrates. After that the CFP-YFP system was used to sort peptides and identify highly sensitive and specific substrate sequences for hK14. We showed that besides its predictable trypsin-like activity, hK14 also possesses a surprising chymotrypsin-like activity. The screening of the Swissprot database was achieved with the most sensitive sequences and allowed the identification of 6 potential human protein substrates for hK14. Three of them, laminin α-5, collagen IV and matrilin-4, which are components of the extracellular matrix were incubated with hK14, by which they were efficiently hydrolyzed. Moreover, electrophoretic separation revealed that degradation of laminin α-5 and matrilin-4 by hK14 generated fragments with identical molecular size than the predicted N-terminal fragments that would result from hK14 specific cleavage, proving the value of phage display substrate to identify potential substrates. Finally, with site-directed mutagenesis, we transformed two well-known serpins (serine protease inhibitors), AAT and ACT (alpha anti-trypsin and alpha anti-chymotrypsin), which inhibit a vast spectrum of human enzymes into highly efficient and specific hK14 inhibitors. These inhibitors will be used to pursue experiments that could help understand hK14 implication in physiological pathways as well as in cancer biology and also to perform their in vivo evalution as potential cancer treatment.