A Survey of Active Learning Algorithms for Supervised Remote Sensing Image Classification

Defining an efficient training set is one of the most delicate phases for the success of remote sensing image classification routines. The complexity of the problem, the limited temporal and financial resources, as well as the high intraclass variance can make an algorithm fail if it is trained with a suboptimal dataset. Active learning aims at building efficient training sets by iteratively improving the model performance through sampling. A user-defined heuristic ranks the unlabeled pixels according to a function of the uncertainty of their class membership and then the user is asked to provide labels for the most uncertain pixels. This paper reviews and tests the main families of active learning algorithms: committee, large margin, and posterior probability-based. For each of them, the most recent advances in the remote sensing community are discussed and some heuristics are detailed and tested. Several challenging remote sensing scenarios are considered, including very high spatial resolution and hyperspectral image classification. Finally, guidelines for choosing the good architecture are provided for new and/or unexperienced user.

Identification of two steroid-responsive promoters of different strength controlled by the same estrogen-responsive element in the 5'-end region of the Xenopus laevis vitellogenin gene A1.

A structural and functional analysis of the 5'-end region of the Xenopus laevis vitellogenin gene A1 revealed two transcription initiation sites located 1.8 kilobases apart. A RNA polymerase II binding assay indicates that both promoters form initiation complexes efficiently. In vitro, using a transcription assay derived from a HeLa whole-cell extract, the upstream promoter is more than 10-fold stronger than the downstream one. In contrast, both promoters have a similar strength in a HeLa nuclear extract. In vivo, that is in estrogen-stimulated hepatocytes, it is the downstream promoter homologous to the one used by the other members of the vitellogenin gene family, which is 50-fold stronger than the upstream promoter. Thus, if functional vitellogenin mRNA results from this latter activity, it would contribute less than 1% to the synthesis of vitellogenin by fully induced Xenopus hepatocytes expressing the four vitellogenin genes. In contrast, both gene A1 promoters are silent in uninduced hepatocytes. Transfection experiments using the Xenopus cell line B3.2 in which estrogen-responsiveness has been introduced reveal that the strong downstream promoter is controlled by an estrogen responsive element (ERE) located 330 bp upstream of it. The upstream promoter can also be controlled by the same ERE. Since the region comprising the upstream promoter is flanked by a 200 base pair long inverted repeat with stretches of homology to other regions of the X. laevis genome, we speculate that it might have been inserted upstream of the vitellogenin gene A1 by a recombination event and consequently brought under control of the ERE lying 1.5 kilobases downstream.

Expression of human estrogen receptor mutants in Xenopus oocytes: correlation between transcriptional activity and ability to form protein-DNA complexes.

The human estrogen receptor (hER) is a trans-acting regulatory protein composed of a series of discrete functional domains. We have microinjected an hER expression vector (HEO) into Xenopus oocyte nuclei and demonstrate, using Western blot assay, that the hER is synthesized. When nuclear extracts from oocytes were prepared and incubated in the presence of a 2.7 kb DNA fragment comprising the 5' end of the vitellogenin gene B2, formation of estrogen-dependent complexes could be visualized by electron microscopy over the estrogen responsive element (ERE). Of crucial importance is the observation that the complex formation is inhibited by the estrogen antagonist tamoxifen, is restored by the addition of the hormone and does not take place with extracts from control oocytes injected with the expression vector lacking the sequences encoding the receptor. The presence of the biologically active hER is confirmed in co-injection experiments, in which HEO is co-introduced with a CAT reporter gene under the control of a vitellogenin promoter containing or lacking the ERE. CAT assays and primer extensions analyses reveal that both the receptor and the ERE are essential for estrogen induced stimulation of transcription. The same approach was used to analyze selective hER mutants. We find that the DNA binding domain (region C) is essential for protein--DNA complex formation at the ERE but is not sufficient by itself to activate transcription from the reporter gene. In addition to region C, both the hormone binding (region E) and amino terminal (region A/B) domains are needed for an efficient transcription activation.(ABSTRACT TRUNCATED AT 250 WORDS)

RNA-driven cyclin-dependent kinase regulation: when CDK9/cyclin T subunits of P-TEFb meet their ribonucleoprotein partners.

The positive transcription elongation factor (P-TEFb) consists of CDK9, a cyclin-dependent kinase and its cyclin T partner. It is required for transcription of most class II genes. Its activity is regulated by non-coding RNAs. The 7SK cellular RNA turns the HEXIM cellular protein into a P-TEFb inhibitor that binds its cyclin T subunit. Thus, P-TEFb activity responds to variations in global cellular transcriptional activity and to physiological conditions linked to cell differentiation, proliferation or cardiac hypertrophy. In contrast, the Tat activation region RNA plays an activating role. This feature at the 5' end of the human immunodeficiency (HIV) viral transcript associates with the viral protein Tat that in turn binds cyclin T1 and recruits active P-TEFb to the HIV promoter. This results in enhanced P-TEFb activity, which is critical for an efficient production of viral transcripts. Although discovered recently, the regulation of P-TEFb becomes a paradigm for non-coding RNAs that regulate transcription factors. It is also a unique example of RNA-driven regulation of a cyclindependent kinase.

Transcriptional activation of the utrophin promoter B by a constitutively active Ets-transcription factor.

Duchenne muscular dystrophy is an X-linked genetic disease caused by the absence of functional dystrophin. Pharmacological upregulation of utrophin, the autosomal homologue of dystrophin, offers a potential therapeutic approach to treat Duchenne patients. Full-length utrophin mRNA is transcribed from two alternative promoters, called A and B. In contrast to the utrophin promoter A, little is known about the factors regulating the activity of the utrophin promoter B. Computer analysis of this second promoter revealed the presence of several conserved binding motives for Ets-transcription factors. Using electrotransfer of cDNA into mouse muscles, we demonstrate that a genetically modified beta-subunit of the Ets-transcription factor GA-binding protein potently activates a utrophin promoter B reporter construct in innervated muscle fibers in vivo. These results make the GA-binding protein and the signaling cascade regulating its activity in muscle cells, potential targets for the pharmacological modulation of utrophin expression in Duchenne patients.

A gene-rich, transcriptionally active environment and the pre-deposition of repressive marks are predictive of susceptibility to KRAB/KAP1-mediated silencing.

AbstractBACKGROUND: KRAB-ZFPs (Krüppel-associated box domain-zinc finger proteins) are vertebrate-restricted transcriptional repressors encoded in the hundreds by the mouse and human genomes. They act via an essential cofactor, KAP1, which recruits effectors responsible for the formation of facultative heterochromatin. We have recently shown that KRAB/KAP1 can mediate long-range transcriptional repression through heterochromatin spreading, but also demonstrated that this process is at times countered by endogenous influences.METHOD: To investigate this issue further we used an ectopic KRAB-based repressor. This system allowed us to tether KRAB/KAP1 to hundreds of euchromatic sites within genes, and to record its impact on gene expression. We then correlated this KRAB/KAP1-mediated transcriptional effect to pre-existing genomic and chromatin structures to identify specific characteristics making a gene susceptible to repression.RESULTS: We found that genes that were susceptible to KRAB/KAP1-mediated silencing carried higher levels of repressive histone marks both at the promoter and over the transcribed region than genes that were insensitive. In parallel, we found a high enrichment in euchromatic marks within both the close and more distant environment of these genes.CONCLUSION: Together, these data indicate that high levels of gene activity in the genomic environment and the pre-deposition of repressive histone marks within a gene increase its susceptibility to KRAB/KAP1-mediated repression.

40Ar/39Ar dating of Penninic Front tectonic displacement (W Alps) during the Lower Oligocene (31-34 Ma)

Direct absolute dating of the Penninic Frontal Thrust tectonic motion is achieved using the Ar-40/Ar-39 technique in the Pelvoux Crystalline Massif (Western Alps). The dated phengites were formed syn-kinematically in shear zones. They underline the brittle-ductile stretching lineation, pressure-shadow fibres and slickensides consistent with underthrusting of the European continental slab below the propagating Penninic Thrust. Chlorite-phengite thermobarometry yields 10-15 km and T similar to 280 degrees C, while Ar-40/Ar-39 phengite ages mainly range between 34 and 30 Ma, with one younger age at 27 Ma. This Early Oligocene age range matches a major tectonic rearrangement of the Alpine chain. Preservation of prograde Ar-40/Ar-39 ages is ascribed to passive exhumation of the Pelvoux shear zone network, sandwiched between more external thrusts and the Penninic Front reactivated as an E-dipping detachment fault. Partial resetting in the Low Temperature part of argon spectra below 24 Ma is ascribed to brittle deformation and alteration of phengites.

Best practice in the management of mild-to-moderately active ulcerative colitis and achieving maintenance of remission using mesalazine.

Optimizing treatment goals in ulcerative colitis requires recognizing the needs of patients. It is increasingly recognized that adapting treatment strategies aligned with patient needs can improve patient compliance and consequently minimize relapse rates. Tailoring of treatment strategies can improve not only patient quality of life, and decrease the number harmed by adverse events from more potent drugs, but can also save valuable healthcare costs by avoiding high-cost treatment interventions associated with acute ulcerative colitis. This review will consider several elements of mesalazine management from the patient perspective based on a range of clinical and patient-focused evidence. By highlighting patient preferences in disease management it is envisaged that this review will aid physicians to optimize treatment decisions with the different mesalazine preparations available.

Age-period-cohort analysis of oral cancer mortality in Europe: the end of an epidemic?

Over the last decade, mortality from oral and pharyngeal cancer has been declining in most European countries, but it had been increasing substantially in Hungary, Slovakia and a few other countries of central Europe, reaching rates comparable to those of lung cancer in several western European countries in males. To update trends in oral cancer mortality and further analyse the recent epidemic in central Europe, official death certifications for oral and pharyngeal cancer for 37 European countries were derived over the period 1970-2007, and an age-period-cohort model was fitted for selected countries. Male oral cancer mortality continued to decline in most European countries, including the Russian Federation, and, more importantly, it also started to decline in some of the countries with the highest male rates, i.e. Hungary and Slovakia; persisting rises were, however, observed in Belarus, Bulgaria and Romania. Oral cancer mortality rates for women were lower than in men and showed no appreciable trend over recent periods in the EU overall. Estimates from the age-period-cohort analysis for most selected countries showed a fall in effects for the cohorts born after the 1950s. For the period effect displayed a rise for the earlier periods, an inversion in the 1990s and a continuous fall up to the last studied period. Only some former non-market economy countries, like Romania, Ukraine and Lithuania, had rising cohort effect trends up to most recent generations. The major finding of this updated analysis of oral cancer mortality is the leveling of the epidemic for men in most European countries, including Hungary and other central European countries, where mortality from this cancer was exceedingly high. These trends essentially reflect the changes in alcohol and tobacco consumption in various populations.

Characteristic bimodal profiles of RNA polymerase II at thousands of active mammalian promoters.

BACKGROUND: In mammals, ChIP-seq studies of RNA polymerase II (PolII) occupancy have been performed to reveal how recruitment, initiation and pausing of PolII may control transcription rates, but the focus is rarely on obtaining finely resolved profiles that can portray the progression of PolII through sequential promoter states. RESULTS: Here, we analyze PolII binding profiles from high-coverage ChIP-seq on promoters of actively transcribed genes in mouse and humans. We show that the enrichment of PolII near transcription start sites exhibits a stereotypical bimodal structure, with one peak near active transcription start sites and a second peak 110 base pairs downstream from the first. Using an empirical model that reliably quantifies the spatial PolII signal, gene by gene, we show that the first PolII peak allows for refined positioning of transcription start sites, which is corroborated by mRNA sequencing. This bimodal signature is found both in mouse and humans. Analysis of the pausing-related factors NELF and DSIF suggests that the downstream peak reflects widespread pausing at the +1 nucleosome barrier. Several features of the bimodal pattern are correlated with sequence features such as CpG content and TATA boxes, as well as the histone mark H3K4me3. CONCLUSIONS: We thus show how high coverage DNA sequencing experiments can reveal as-yet unnoticed bimodal spatial features of PolII accumulation that are frequent at individual mammalian genes and reminiscent of transcription initiation and pausing. The initiation-pausing hypothesis is corroborated by evidence from run-on sequencing and immunoprecipitation in other cell types and species.

Circulating soluble CR1 (CD35). Serum levels in diseases and evidence for its release by human leukocytes.

C receptor type 1 (CR1, CD35) is present in a soluble form in plasma (sCR1). Soluble CR1 was measured with a specific ELISA assay in normal individuals and in patients with different diseases. The mean serum concentration of sCR1 in 31 normal donors was 31.4 +/- 7.8 ng/ml, and was identical in plasma. An increase in sCR1 was observed in 36 patients with end-stage renal failure on dialysis (54.8 +/- 11.7 ng/ml, p < 0.0001), and in 22 patients with liver cirrhosis (158.3 +/- 49.9 ng/ml, p < 0.0001). The mean sCR1 levels dropped from 181 +/- 62.7 to 52.1 +/- 24.0 ng/ml (p < 0.001) in nine patients who underwent liver transplantation, and was 33.5 +/- 7.3 in 10 patients with functioning renal grafts, indicating that the increase in sCR1 was reversible. Soluble CR1 was elevated in some hematologic malignancies (> 47 ng/ml), which included B cell lymphoma (12/19 patients), Hodgkin's lymphoma (4/4), and chronic myeloproliferative syndromes (4/5). By contrast, no increase was observed in acute myeloid or lymphoblastic leukemia (10) or myeloma (5). In two patients with chronic myeloproliferative syndromes, sCR1 decreased rapidly after chemotherapy. The mean concentration of sCR1 was not significantly modified in 181 HIV-infected patients at various stages of the disease (34.8 +/- 14.4 ng/ml), and in 13 patients with active SLE (38.3 +/- 19.6 ng/ml), although in both groups the number of CR1 was diminished on E. There was a weak but significant correlation between sCR1 and CR1 per E in HIV infection and SLE (r = 0.39, p < 0.0001, and r = 0.60, p < 0.03 respectively). In vitro, monocytes, lymphocytes, and neutrophils were found to release sCR1 into culture supernatants. In vivo, sCR1 was detected in the serum of SCID mice populated with human peripheral blood leukocytes. The sCR1 levels correlated with those of human IgG (r = 0.97, p < 0.0001), suggesting synthesis of sCR1 by the transferred lymphocytes. The mechanisms underlining the increased levels of sCR1 and its biologic consequences remain to be defined.