SEARCH · Search PubMed

Results for “variational inference”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 289 records · Page 16Linked to original sources

PubMed2006

Blind deconvolution using a variational approach to parameter, image, and blur estimation.

Following the hierarchical Bayesian framework for blind deconvolution problems, in this paper, we propose the use of simultaneous autoregressions as prior distributions for both the image and blur, and gamma distributions for the unknown parameters (hyperparameters) of the priors and the image formation noise. We show how the gamma distributions on the unknown hyperparameters can be used to prevent the proposed blind deconvolution method from converging to undesirable image and blur estimates and also how these distributions can be inferred in realistic situations. We apply variational methods to approximate the posterior probability of the unknown image, blur, and hyperparameters and propose two different approximations of the posterior distribution. One of these approximations coincides with a classical blind deconvolution method. The proposed algorithms are tested experimentally and compared with existing blind deconvolution methods.

Algorithms↗

PubMed2004

Genetic diversity of peanut (Arachis hypogaea L.) and its wild relatives based on the analysis of hypervariable regions of the genome.

BACKGROUND: The genus Arachis is native to a region that includes Central Brazil and neighboring countries. Little is known about the genetic variability of the Brazilian cultivated peanut (Arachis hypogaea, genome AABB) germplasm collection at the DNA level. The understanding of the genetic diversity of cultivated and wild species of peanut (Arachis spp.) is essential to develop strategies of collection, conservation and use of the germplasm in variety development. The identity of the ancestor progenitor species of cultivated peanut has also been of great interest. Several species have been suggested as putative AA and BB genome donors to allotetraploid A. hypogaea. Microsatellite or SSR (Simple Sequence Repeat) markers are co-dominant, multiallelic, and highly polymorphic genetic markers, appropriate for genetic diversity studies. Microsatellite markers may also, to some extent, support phylogenetic inferences. Here we report the use of a set of microsatellite markers, including newly developed ones, for phylogenetic inferences and the analysis of genetic variation of accessions of A. hypogea and its wild relatives. RESULTS: A total of 67 new microsatellite markers (mainly TTG motif) were developed for Arachis. Only three of these markers, however, were polymorphic in cultivated peanut. These three new markers plus five other markers characterized previously were evaluated for number of alleles per locus and gene diversity using 60 accessions of A. hypogaea. Genetic relationships among these 60 accessions and a sample of 36 wild accessions representative of section Arachis were estimated using allelic variation observed in a selected set of 12 SSR markers. Results showed that the Brazilian peanut germplasm collection has considerable levels of genetic diversity detected by SSR markers. Similarity groups for A. hypogaea accessions were established, which is a useful criteria for selecting parental plants for crop improvement. Microsatellite marker transferability was up to 76% for species of the section Arachis, but only 45% for species from the other eight Arachis sections tested. A new marker (Ah-041) presented a 100% transferability and could be used to classify the peanut accessions in AA and non-AA genome carriers. CONCLUSION: The level of polymorphism observed among accessions of A. hypogaea analyzed with newly developed microsatellite markers was low, corroborating the accumulated data which show that cultivated peanut presents a relatively reduced variation at the DNA level. A selected panel of SSR markers allowed the classification of A. hypogaea accessions into two major groups. The identification of similarity groups will be useful for the selection of parental plants to be used in breeding programs. Marker transferability is relatively high between accessions of section Arachis. The possibility of using microsatellite markers developed for one species in genetic evaluation of other species greatly reduces the cost of the analysis, since the development of microsatellite markers is still expensive and time consuming. The SSR markers developed in this study could be very useful for genetic analysis of wild species of Arachis, including comparative genome mapping, population genetic structure and phylogenetic inferences among species.

Alleles↗

PubMed2005

No association between insulin gene variation and adult metabolic phenotypes in a large Finnish birth cohort.

AIMS/HYPOTHESIS: Although the variable number tandem repeat (VNTR) minisatellite 5' to the insulin gene is among the most studied polymorphisms in diabetes, the relationships between VNTR variation, diabetes-related traits and predisposition to type 2 diabetes remain unclear. Since inadequate sample size is likely to have been an obstacle to reliable inference, we examined the relationship between VNTR variation and a range of diabetes-related traits in a cohort of 5,753 Finnish adults. MATERIALS AND METHODS: VNTR genotypes were derived, by typing at the -23HphI variant site, for 5,646 individuals from the Northern Finland Birth Cohort 1966. Associations were sought between these genotypes and a range of anthropometric (BMI, WHR), physiological (blood pressure) and biochemical (fasting glucose, insulin, lipids, indices of insulin sensitivity and beta cell function) measures obtained at clinical examination at 31 years. RESULTS: We found no evidence that VNTR genotype was significantly associated with measures of insulin secretion, insulin sensitivity, glycaemia, adiposity or blood pressure. CONCLUSIONS/INTERPRETATION: Despite evidence from several relatively small studies suggesting that INS-VNTR genotypes are associated with predisposition to type 2 diabetes, reduced beta cell function and measures of adiposity, the present study failed to detect any association with a range of diabetes-related traits. Taken with other recent studies in large population-based cohorts, these data suggest that previous studies have, at the very least, overestimated the influence of the INS-VNTR on type 2 diabetes-related traits. The effects of INS-VNTR variation on insulin transcription observed in vitro appear not to translate into detectable differences in basal insulin secretion in humans.

Adult↗

PubMed2002

Comparative methods and evolution.

Comparative biology provides a framework for the study of evolution, by seeking answers to the question of why traits evolved. However, the difficulties of making inferences about the biological causes of trait variation and covariation resulted in the development of very different approaches to comparative analyses, which are outlined and contrasted here. The Homology approach differs from all other approaches as it attempts to explain historical uniques, whereas the various homoplasy approaches draw inferences from repeated origination of correlated traits. A distinction is made between those approaches deriving information from correlated transitions of characters on a cladogram (Homoplasy I) and those that simply extract correlated differences between pairs of phylogenetically independent taxa (Homoplasy II). The latter approach has been implemented in well-developed analytical procedures ("comparative methods") which, however, are principally non-historical and thus their inferences about evolution are indirect. In contrast, the study of correlation of character transitions seeks to explain trait variation as the outcome of evolution, but is hampered by the difficulty of reconstructing ancestral character states, a prerequisite for the analysis of correlated transitions. Model-based techniques for ancestral state reconstruction are being developed, but the lack of data contained in single characters limits the biological reality of parameter-intensive models.

Animals↗

PubMed2004

Distribution of genetic variation in the growth hormone 1 gene in Atlantic salmon (Salmo salar) populations from Europe and North America.

The level and hierarchical distribution of genetic variation in complete sequences of the Atlantic salmon (Salmo salar) growth hormone (GH1) gene were investigated in populations from Europe and North America with a view to inferring the major evolutionary forces affecting genetic variation at this locus. Seventeen polymorphic sites were identified in complete sequences from nine populations, with levels of noncoding (intron and untranslated region sequences) nucleotide diversity being similar to those observed in other species. No variation, however, was observed in exonic sequences, indicating that nucleotide diversity in the Atlantic salmon GH1 gene is three and 25 times less than that estimated for human and Drosophila coding sequences, respectively. This suggests that purifying selection is the predominant contemporary force controlling the molecular evolution of GH1 coding sequences. Comparison of haplotype relationships within and between populations indicated that differentiation between populations from Europe and North America was greater than within-continent comparisons. However, several haplotypes observed in the northernmost European populations were more similar to those observed in North American than to any other haplotypes observed in Europe. This is most likely to be a result of historical, rather than contemporary, gene flow. Neutrality test statistics, such as Tajima's D, were significantly positive in the European populations in which North American-like haplotypes were observed. Although a positive Tajima's D is commonly interpreted as the signal of balancing selection, a more likely explanation in this case is that either historical migration or ascertainment bias, rather than within population local adaptation, has given rise to an excess of intermediate frequency alleles.

Amino Acid Sequence↗

PubMed1984

Intracranial pressure variations associated with activation of the cholinoceptive pontine inhibitory area in the unanesthetized drug-free cat.

Intracranial pressure (ICP) was recorded continuously in chronically prepared, unanesthetized cats in order to investigate the effects on ICP of the cholinergic agonist, carbamylcholine (carbachol), injected by microsyringe needles into the dorsal pontine tegmentum. As reported previously, carbachol microinjections into the medial part of the cholinoceptive pontine inhibitory area (CPIA) located ventromedially to the locus coeruleus produced a comatose state characterized by a profound unresponsiveness to external stimuli, desynchronized electroencephalograms (EEG's), and suppression of postural somatomotor and sympathetic visceromotor functions. Four of six ICP records following carbachol microinjections into the CPIA showed small but significant increases which occurred in association with these carbachol effects. Tracings of ICP increases ranged up to 3.2 mm Hg and were similar in shape to plateau waves. The start and resolution of these carbachol-induced ICP variations were closely associated with the onset and termination of EEG desynchronization and signs of reduced cervical sympathetic tone, but not with changes in systemic arterial blood pressure or arterial pCO2. Temporal associations between ICP increases, desynchronized EEG's, and signs of reduced sympathetic tone were repeatedly confirmed during recovery periods associated both with recurrent comatose states following wakefulness produced by various intensities of external stimulation and with spontaneously occurring states resembling rapid eye movement sleep. The authors infer that carbachol-induced ICP variations may be produced by increased cerebral blood volume in response to accelerated cerebral metabolism and reduced vasoconstrictor tone of cervical sympathetic nerves. The simultaneous occurrence of continuously accelerated cerebral metabolism and reduced cervical sympathetic tone can neither be seen in physiologically normal, awake organisms nor produced by other known experimental manipulations of the central nervous system. Such a paradoxical relationship appears to be a unique consequence of activity within the CPIA. These data suggest that episodic activity within the CPIA may provide at least one endogenous neural basis for plateau waves seen during certain pathological conditions such as disturbed cerebrospinal fluid (CSF) absorption or with reduced equilibrium volume of CSF space.

Animals↗

PubMed1999

Extensive intraspecific chloroplast DNA (cpDNA) variation in the alpine Draba aizoides L. (Brassicaceae): haplotype relationships and population structure.

Chloroplast DNA (cpDNA) sequence variation is currently the most widely used tool for the inference of phylogenetic relationships among plants at all taxonomic levels. Generally, noncoding regions tend to evolve faster than coding sequences and have recently been applied to the study of phylogenetic relationships among closely related taxa. An implicit assumption of many of these studies is that intraspecific cpDNA variation is either absent or low and therefore will not interfere with the reconstruction of interspecific relationships. A survey of cpDNA sequence variation in the common alpine plant species Draba aizoides L. was undertaken to assess levels of intraspecific cpDNA sequence variation. These levels were compared to levels of interspecific sequence divergence between D. aizoides and related alpine Draba species. Intraspecific cpDNA sequence divergence was extensive in D. aizoides, and intraspecific differences were often larger than interspecific differences. cpDNA haplotype relationships were explored using a maximum parsimony approach and minimum-spanning networks. Results from both methods were largely congruent but comparisons provided interesting insights into the presumed evolutionary history of cpDNA haplotypes. A combined effect of cpDNA introgression and complex lineage sorting was inferred to explain the pattern of cpDNA variation found in D. aizoides. Our results suggest that intraspecific cpDNA variation can be extensive and that intraspecific variation needs to be taken into account when inferring phylogenetic relationships among closely related taxa.

Brassicaceae↗

PubMed1999

Origin of the unisexual lizard Gymnophthalmus underwoodi (Gymnophthalmidae) inferred from mitochondrial DNA nucleotide sequences.

Analysis of variation in nucleotide sequences in mitochondrial DNA genes among species of Gymnophthalmus supports three major conclusions. First, samples of G. underwoodi, a unisexual species of hybrid origin, have identical nucleotide sequences, supporting the hypothesis that all individuals from localities in the Guianan Region of South America and the Lesser Antilles descended from a single hybridization event. Second, G. underwoodi and G. cryptus share derived similarities in nucleotide sequences of the maternally inherited mitochondrial DNA, suggesting that the female participant in the hybridization event that produced the hybrid species was either G. cryptus or most closely related to it. Last, phylogenetic analyses recovered historical structure among population samples of G. speciosus supporting previous claims that there is unrecognized diversity under that binomial, and some analyses suggest that G. speciosus does not comprise a natural group.

Animals↗

PubMed1991

Geographic origin of the US and Brazilian Aedes albopictus inferred from allozyme analysis.

A survey of genetic variation using allozymes was conducted on 57 populations of Aedes albopictus, an Asian mosquito that was introduced into the US and Brazil in the mid-1980s. Our objective was to quantify the patterns of genetic variation among its populations and to use that information to trace genetically the geographic origin of the US and Brazilian populations. Populations from the various regions were genetically distinct from one another. Populations from within a region were genetically similar to one another, which resulted in a significant positive correlation between geographic and genetic distances. A discriminant analysis of allele frequencies separated populations from the various countries into nine non-overlapping clusters; the US, Japanese, Chinese and Brazilian populations formed closely placed, but distinct, clusters. The probability of assigning a population to the correct country was 98 per cent. The US and the Brazilian populations were closest in terms of genetic distance from the Japanese populations. Based on discriminant and genetic distance analyses, we conclude that the US and the Brazilian Aedes albopictus originated in Japan.

Aedes↗

PubMed1993

Genetic variation and phylogenetic relationships among worldwide collections of the Russian wheat aphid, Diuraphis noxia (Mordvilko), inferred from allozyme and RAPD-PCR markers.

Genetic analyses were conducted on Diuraphis noxia (Mordvilko) populations collected from wheat, barley and other grasses from various countries throughout the world. These collections had been found to contain clones that differed in virulence from various cultivars, cuticular hydrocarbon profiles and life cycle characters. Discrete genetic markers analysed in this study included allozymes and arbitrary regions of the genome amplified by the polymerase chain reaction (RAPD-PCR). In all, 23 enzymes were evaluated; 17 of these enzymes representing 20 isozyme loci, were judged suitable for allozyme analysis. Polymorphisms were detected at three (15 per cent) loci: beta-esterase (beta-EST), phosphoglucose isomerase (PGI), and 6-phosphogluconate dehydrogenase (6-PGDH). The average expected heterozygosity amongst these loci was 4.9 per cent in the worldwide collection. Allozyme variation was absent within most populations, particularly within those countries where the species was recently introduced. Much greater genetic variation was detected when populations were analysed with RAPD-PCR. Populations were analysed with 69 polymorphic bands amplified by seven primers. All populations could be distinguished with this method. Cluster analyses indicated strong similarities between U.S.A. populations and collections from South Africa, Mexico, France and Turkey. The most variation was detected among populations from the Middle East and southern Russia.

Animals↗

PubMed1982

Diurnal variation of cholesterol precursors squalene and methyl sterols in human plasma lipoproteins.

Animal cholesterol synthesis shows a marked diurnal variation, a phenomenon, at the moment, not known to occur in man. Since cholesterol precursors in serum reflect overall cholesterol synthesis in many conditions, a 24-hr profile of squalene and methyl sterols was studied in plasma lipoproteins in order to demonstrate whether these cholesterol precursors could exhibit a diurnal cycling in healthy human subjects. During the 24-hr period, lipoproteins of density < 1.006 g/ml transported 30-50% of the plasma squalene. Free methyl sterols were found mainly in low density lipoproteins (LDL) and esterified methyl sterols in LDL and high density lipoproteins (HDL). Postprandial hyperlipidemia at noon was associated with an inconsistent increase of the squalene and free methyl sterol concentrations in the lipoproteins of density < 1.006 g/ml. In terms of micro g per mg of cholesterol, the precursor contents were, however, low in each lipoprotein during the daytime. During the night and early morning, the values were several times higher. Thus the peak plasma squalene and methyl sterol contents occurred at midnight and 4 am. The highest variation was found for squalene in the density class < 1.006 g/ml and for lanosterol and diunsaturated dimethyl sterol in LDL and HDL. For different methyl sterols, the mean diurnal variation was 3.5- to 6.9-fold in LDL, 2.0- to 4.5-fold in HDL, and 2.6- to 3.6-fold in the density class < 1.006 g/ml. The respective values for squalene were 2.2, 1.4, and 2.9. Esterified methyl sterols varied slightly in the density class < 1.006 g/ml only, and the percentage esterification exhibited a diurnal fluctuation that was the reciprocal of that of free methyl sterol levels. The rapid and marked diurnal fluctuation of squalene and free methyl sterols in plasma lipoproteins suggests that these precursors are metabolized on and off lipoproteins. The variation is most likely caused by changes in cholesterol synthesis, inferring that circadian rhythm also regulates human cholesterol production.-Miettinen, T. A. Diurnal variation of cholesterol precursors squalene and methyl sterols in human plasma lipoproteins.

Adolescent↗

PubMed1997

Genetic structure of an aphid studied using microsatellites: cyclic parthenogenesis, differentiated lineages and host specialization.

In a previous study, samples of the grain aphid Sitobion avenae (F.) were collected from wheat and adjacent cocksfoot hosts in a population thought to be primarily parthenogenetic, and DNA from individual aphids was analysed with a multilocus technique. Here we have applied single-locus microsatellites and a mitochondrial DNA marker to a subset of the same DNA extracts, and have made several additional inferences about important genetic and population processes in S. avenae. Microsatellite analysis indicated very high levels of genic and genotypic variation. S. avenae fell into three genotypic groups inferred to be almost noninterbreeding, while analysis of linkage and Hardy-Weinberg equilibria suggested high levels of sexual recombination within each genotypic group. Host specialization was evident: one lineage was found only on wheat, and one (bearing many alleles inferred to be introgressed from the blackberry-grass aphid S. fragariae (Walker)) was found only on cocksfoot. The third group of interrelated genotypes was found commonly on both hosts. Although most genotypes were found only once, some were much more numerous in the sample than expected from the frequency of the alleles they contained. This, and rapid temporal changes in genotypic composition of samples, indicates strong selective differences between genotypes and lineages. In the major genotypic group, the commonest genotypes were significantly more homozygous than were rare ones: thus these data may help to explain the frequent observation of homozygous excess in aphid allozymes. The genotype group showing S. avenae-like as well as S. fragariae-like alleles also carried S. fragariae-like mitochondrial DNA in at least 25/31 cases, indicating gender-asymmetrical hybridization.

Animals↗

PubMed2007

Direct and indirect inferences on parasite mating and gene transmission patterns. Pangamy in the cattle tick Rhipicephalus (Boophilus) microplus.

Mating system plays a determinant role in the maintenance and distribution of genetic variation. Difficulties in applying standard methods of indirect inferences onto parasitic life-cycles partly explain the current lack of knowledge on parasite mating systems. The present study develops a combination of direct and indirect inference methods circumventing such difficulties, and illustrates in particular how such a combination modifies our understanding of the biology of the southern cattle tick, Rhipicephalus microplus. To directly assess how females and males pair in natural populations, we sampled 225 mating pairs, genotyped them at six microsatellite loci, and analysed the correlation between mating status and genetic relatedness. This analysis revealed pangamy, i.e. a random association of male and female genotypes. In addition, indirect methods inferred that sib-groups of ticks exploit distinct individual-cows, and hence that ticks probably move among cows during their parasitic lifetime. Altogether, these results negate the expectation of a high frequency of sib-mating, and show the coexistence of genetically diverging stocks within tick cohorts and populations. These results have several consequences for our understanding of R. microplus. For instance, while existing models of pesticide resistance management look perfectly applicable to this species, the epidemiological models of the micropathogens it vectors needs revision. More largely, the methods developed here would help clarify the evolutionary patterns of any dioceous parasite.

Animals↗

PubMed2000

Phylogeography and natural selection in the Tenerife gecko Tarentola delalandii: testing historical and adaptive hypotheses.

Combining phylogeographic and matrix correspondence approaches in the analysis of geographical variation provides a fruitful approach to inferring the causes of molecular and morphological evolution within species. Here we present a study on the gecko Tarentola delalandii on the island of Tenerife, Canary Islands, which provides an outstanding model of an exceptionally high degree of phylogeographic differentiation in magnitude and pattern on a small spatial scale. We reconstruct the population history of T. delalandii using phylogeographic information, matrix correspondence tests and estimates of divergence times in conjunction with geological data. It appears that populations differentiated on three precursor islands and secondary contact followed the junction of these islands. The cytochrome b sequence appears to be evolving at least at approximately 1% per million years in this species. Matrix correspondence tests indicate that morphological character systems may reflect ecological selection regimes (colour pattern), history (body dimensions) or both (scalation). The results imply that natural selection can override a historical legacy, but also underline the potential relevance of molecular phylogenetic data for the interpretation of geographical variation in morphology.

Animals↗

PubMed2026

Comparative Population Genomics of Relictual Caribbean Island Gossypium hirsutum.

Gossypium hirsutum is the world's most important source of cotton fibre, yet the diversity and population structure of its wild forms remain largely unexplored. The complex domestication history of G. hirsutum combined with reciprocal introgression with a second domesticated species, G. barbadense, has generated a wealth of morphological forms and feral derivatives of both species and their interspecies recombinants, which collectively are scattered across a large geographic range in arid regions of the Caribbean basin. Here we assessed genetic diversity within and among populations from two Caribbean islands, Puerto Rico (n = 43, five sites) and Guadeloupe (n = 25, one site), which contain putative wild or introgressed forms. Using whole-genome resequencing data and a phylogenomic framework derived from a broader genomic survey, we parsed individuals into feral derivatives and truly wild forms. Feral cottons display uneven levels of genetic and morphological resemblance to domesticated cottons, with diverse patterns of genetic variation and heterozygosity. These patterns are inferred to reflect a complex history of interspecific and intraspecific gene flow that is spatially highly variable in its effects. Wild cottons in both Caribbean islands appear to be relatively inbred, especially the Guadeloupe samples. Our results highlight the dynamics of population demographics in relictual wild cottons that experienced profound genetic bottlenecks associated with repeated habitat destruction superimposed on a natural ecogeographical distribution comprising widely scattered populations. These results have implications for conservation and utilisation of wild diversity in G. hirsutum.

Genetics, Population↗

PubMed2004

Haplotype diversity across 100 candidate genes for inflammation, lipid metabolism, and blood pressure regulation in two populations.

Recent studies have suggested that a significant fraction of the human genome is contained in blocks of strong linkage disequilibrium, ranging from ~5 to >100 kb in length, and that within these blocks a few common haplotypes may account for >90% of the observed haplotypes. Furthermore, previous studies have suggested that common haplotypes in candidate genes are generally shared across populations and represent the majority of chromosomes in each population. The conclusions drawn from these preliminary studies, however, are based on an incomplete knowledge of the variation in the regions examined. To bridge this gap in knowledge, we have completely resequenced 100 candidate genes in a population of African descent and one of European descent. Although these genes have been well studied because of their medical importance, we demonstrate that a large amount of sequence variation has not yet been described. We also report that the average number of inferred haplotypes per gene, when complete data is used, is higher than in previous reports and that the number and proportion of all haplotypes represented by common haplotypes per gene is variable. Furthermore, we demonstrate that haplotypes shared between the two populations constitute only a fraction of the total number of haplotypes observed and that these shared haplotypes represent fewer of the African-descent chromosomes than was expected from previous studies. Finally, we show that restricting variation discovery to coding regions does not adequately describe all common haplotypes or the true haplotype block structure observed when all common variation is used to infer haplotypes. These data, derived from complete knowledge of genetic variation in these genes, suggest that the haplotype architecture of candidate genes across the human genome is more complex than previously suggested, with important implications for candidate gene and genomewide association studies.

Africa↗

PubMed2003

The role of occlusion in the perception of depth, lightness, and opacity.

A theory is presented that explains how the visual system infers the lightness, opacity, and depth of surfaces from stereoscopic images. It is shown that the polarity and magnitude of image contrast play distinct roles in surface perception, which can be captured by 2 principles of perceptual inference. First, a contrast depth asymmetry principle articulates how the visual system computes the ordinal depth and lightness relationships from the polarity of local, binocularly matched image contrast. Second, a global transmittance anchoring principle expresses how variations in contrast magnitudes are used to infer the presence of transparent surfaces. It is argued that these principles provide a unified explanation of how the visual system computes the 3-D surface structure of opaque and transparent surfaces.

Depth Perception↗

PubMed2006

Flexible empirical Bayes models for differential gene expression.

MOTIVATION: Inference about differential expression is a typical objective when analyzing gene expression data. Recently, Bayesian hierarchical models have become increasingly popular for this type of problem. The two most common hierarchical models are the hierarchical Gamma-Gamma (GG) and Lognormal-Normal (LNN) models. However, to facilitate inference, some unrealistic assumptions have been made. One such assumption is that of a common coefficient of variation across genes, which can adversely affect the resulting inference. RESULTS: In this paper, we extend both the GG and LNN modeling frameworks to allow for gene-specific variances and propose EM based algorithms for parameter estimation. The proposed methodology is evaluated on three experimental datasets: one cDNA microarray experiment and two Affymetrix spike-in experiments. The two extended models significantly reduce the false positive rate while keeping a high sensitivity when compared to the originals. Finally, using a simulation study we show that the new frameworks are also more robust to model misspecification. AVAILABILITY: The R code for implementing the proposed methodology can be downloaded at http://www.stat.ubc.ca/~c.lo/FEBarrays. SUPPLEMENTARY INFORMATION: The supplementary material is available at http://www.stat.ubc.ca/~c.lo/FEBarrays/supp.pdf.

Algorithms↗