Adaptively weighted maximum likelihood estimation of discrete distributions

SummaryDiscrete data arise in various research fields, typically when the observations are count data.I propose a robust and efficient parametric procedure for estimation of discrete distributions. The estimation is done in two phases. First, a very robust, but possibly inefficient, estimate of the model parameters is computed and used to indentify outliers. Then the outliers are either removed from the sample or given low weights, and a weighted maximum likelihood estimate (WML) is computed.The weights are determined via an adaptive process such that if the data follow the model, then asymptotically no observation is downweighted.I prove that the final estimator inherits the breakdown point of the initial one, and that its influence function at the model is the same as the influence function of the maximum likelihood estimator, which strongly suggests that it is asymptotically fully efficient.The initial estimator is a minimum disparity estimator (MDE). MDEs can be shown to have full asymptotic efficiency, and some MDEs have very high breakdown points and very low bias under contamination. Several initial estimators are considered, and the performances of the WMLs based on each of them are studied.It results that in a great variety of situations the WML substantially improves the initial estimator, both in terms of finite sample mean square error and in terms of bias under contamination. Besides, the performances of the WML are rather stable under a change of the MDE even if the MDEs have very different behaviors.Two examples of application of the WML to real data are considered. In both of them, the necessity for a robust estimator is clear: the maximum likelihood estimator is badly corrupted by the presence of a few outliers.This procedure is particularly natural in the discrete distribution setting, but could be extended to the continuous case, for which a possible procedure is sketched.RésuméLes données discrètes sont présentes dans différents domaines de recherche, en particulier lorsque les observations sont des comptages.Je propose une méthode paramétrique robuste et efficace pour l'estimation de distributions discrètes. L'estimation est faite en deux phases. Tout d'abord, un estimateur très robuste des paramètres du modèle est calculé, et utilisé pour la détection des données aberrantes (outliers). Cet estimateur n'est pas nécessairement efficace. Ensuite, soit les outliers sont retirés de l'échantillon, soit des faibles poids leur sont attribués, et un estimateur du maximum de vraisemblance pondéré (WML) est calculé.Les poids sont déterminés via un processus adaptif, tel qu'asymptotiquement, si les données suivent le modèle, aucune observation n'est dépondérée.Je prouve que le point de rupture de l'estimateur final est au moins aussi élevé que celui de l'estimateur initial, et que sa fonction d'influence au modèle est la même que celle du maximum de vraisemblance, ce qui suggère que cet estimateur est pleinement efficace asymptotiquement.L'estimateur initial est un estimateur de disparité minimale (MDE). Les MDE sont asymptotiquement pleinement efficaces, et certains d'entre eux ont un point de rupture très élevé et un très faible biais sous contamination. J'étudie les performances du WML basé sur différents MDEs.Le résultat est que dans une grande variété de situations le WML améliore largement les performances de l'estimateur initial, autant en terme du carré moyen de l'erreur que du biais sous contamination. De plus, les performances du WML restent assez stables lorsqu'on change l'estimateur initial, même si les différents MDEs ont des comportements très différents.Je considère deux exemples d'application du WML à des données réelles, où la nécessité d'un estimateur robuste est manifeste : l'estimateur du maximum de vraisemblance est fortement corrompu par la présence de quelques outliers.La méthode proposée est particulièrement naturelle dans le cadre des distributions discrètes, mais pourrait être étendue au cas continu.

Cryptic diversity among Western Palearctic tree frogs: postglacial range expansion, range limits, and secondary contacts of three European tree frog lineages (Hyla arborea group).

We characterize divergence times, intraspecific diversity and distributions for recently recognized lineages within the Hyla arborea species group, based on mitochondrial and nuclear sequences from 160 localities spanning its whole distribution. Lineages of H. arborea, H. orientalis, H. molleri have at least Pliocene age, supporting species level divergence. The genetically uniform Iberian H. molleri, although largely isolated by the Pyrenees, is parapatric to H. arborea, with evidence for successful hybridization in a small Aquitanian corridor (southwestern France), where the distribution also overlaps with H. meridionalis. The genetically uniform H. arborea, spread from Crete to Brittany, exhibits molecular signatures of a postglacial range expansion. It meets different mtDNA clades of H. orientalis in NE-Greece, along the Carpathians, and in Poland along the Vistula River (there including hybridization). The East-European H. orientalis is strongly structured genetically. Five geographic mitochondrial clades are recognized, with a molecular signature of postglacial range expansions for the clade that reached the most northern latitudes. Hybridization with H. savignyi is suggested in southwestern Turkey. Thus, cryptic diversity in these Pliocene Hyla lineages covers three extremes: a genetically poor, quasi-Iberian endemic (H. molleri), a more uniform species distributed from the Balkans to Western Europe (H. arborea), and a well-structured Asia Minor-Eastern European species (H. orientalis).

Adaptive divergence of ancient gene duplicates in the avian MHC class II beta.

Gene duplication and neofunctionalization are known to be important processes in the evolution of phenotypic complexity. They account for important evolutionary novelties that confer ecological adaptation, such as the major histocompatibility complex (MHC), a multigene family crucial to the vertebrate immune system. In birds, two MHC class II β (MHCIIβ) exon 3 lineages have been recently characterized, and two hypotheses for the evolutionary history of MHCIIβ lineages were proposed. These lineages could have arisen either by 1) an ancient duplication and subsequent divergence of one paralog or by 2) recent parallel duplications followed by functional convergence. Here, we compiled a data set consisting of 63 MHCIIβ exon 3 sequences from six avian orders to distinguish between these hypotheses and to understand the role of selection in the divergent evolution of the two avian MHCIIβ lineages. Based on phylogenetic reconstructions and simulations, we show that a unique duplication event preceding the major avian radiations gave rise to two ancestral MHCIIβ lineages that were each likely lost once later during avian evolution. Maximum likelihood estimation shows that following the ancestral duplication, positive selection drove a radical shift from basic to acidic amino acid composition of a protein domain facing the α-chain in the MHCII α β-heterodimer. Structural analyses of the MHCII α β-heterodimer highlight that three of these residues are potentially involved in direct interactions with the α-chain, suggesting that the shift following duplication may have been accompanied by coevolution of the interacting α- and β-chains. These results provide new insights into the long-term evolutionary relationships among avian MHC genes and open interesting perspectives for comparative and population genomic studies of avian MHC evolution.

Using a Bayesian method to assign individuals to karyotypic taxa in shrew hybrid zones.

Individuals sampled in hybrid zones are usually analysed according to their sampling locality, morphology, behaviour or karyotype. But the increasing availability of genetic information more and more favours its use for individual sorting purposes and numerous assignment methods based on the genetic composition of individuals have been developed. The shrews of the Sorex araneus group offer good opportunities to test the genetic assignment on individuals identified by their karyotype. Here we explored the potential and efficiency of a Bayesian assignment method combined or not with a reference dataset to study admixture and individual assignment in the difficult context of two hybrid zones between karyotypic species of the Sorex araneus group. As a whole, we assigned more than 80% of the individuals to their respective karyotypic categories (i.e. 'pure' species or hybrids). This assignment level is comparable to what was obtained for the same species away from hybrid zones. Additionally, we showed that the assignment result for several individuals was strongly affected by the inclusion or not of a reference dataset. This highlights the importance of such comparisons when analysing hybrid zones. Finally, differences between the admixture levels detected in both hybrid zones support the hypothesis of an impact of chromosomal rearrangements on gene flow.

Low postseroconversion CD4 count and rapid decrease of CD4 density identify HIV+ fast progressors.

CD4 expression in HIV replication is paradoxical: HIV entry requires high cell-surface CD4 densities, but replication requires CD4 down-modulation. However, is CD4 density in HIV+ patients affected over time? Do changes in CD4 density correlate with disease progression? Here, we examined the role of CD4 density for HIV disease progression by longitudinally quantifying CD4 densities on CD4+ T cells and monocytes of ART-naive HIV+ patients with different disease progression rates. This was a retrospective study. We defined three groups of HIV+ patients by their rate of CD4+ T cell loss, calculated by the time between infection and reaching a CD4 level of 200 cells/microl: fast (<7.5 years), intermediate (7.5-12 years), and slow progressors (>12 years). Mathematical modeling permitted us to determine the maximum CD4+ T cell count after HIV seroconversion (defined as "postseroconversion CD4 count") and longitudinal profiles of CD4 count and density. CD4 densities were quantified on CD4+ T cells and monocytes from these patients and from healthy individuals by flow cytometry. Fast progressors had significantly lower postseroconversion CD4 counts than other progressors. CD4 density on T cells was lower in HIV+ patients than in healthy individuals and decreased more rapidly in fast than in slow progressors. Antiretroviral therapy (ART) did not normalize CD4 density. Thus, postseroconversion CD4 counts define individual HIV disease progression rates that may help to identify patients who might benefit most from early ART. Early discrimination of slow and fast progressors suggests that critical events during primary infection define long-term outcome. A more rapid CD4 density decrease in fast progressors might contribute to progressive functional impairments of the immune response in advanced HIV infection. The lack of an effect of ART on CD4 density implies a persistent dysfunctional immune response by uncontrolled HIV infection.

The truncated mean of an asymmetric distribution

This paper investigates a simple procedure to estimate robustly the mean of an asymmetric distribution. The procedure removes the observations which are larger or smaller than certain limits and takes the arithmetic mean of the remaining observations, the limits being determined with the help of a parametric model, e.g., the Gamma, the Weibull or the Lognormal distribution. The breakdown point, the influence function, the (asymptotic) variance, and the contamination bias of this estimator are explored and compared numerically with those of competing estimates.

Asymmetry of Functional Connectivity in Schizophrenia at Different Spatial Scales

Introduction: The interhemispheric asymmetries that originate from connectivity-related structuring of the cerebral cortex are compromised in schizophrenia (SZ). Recently, we have revealed the whole-head topography of EEG synchronization in SZ (Jalili et al. 2007; Knyazeva et al. 2008). Here we extended the analysis to assess the abnormality in the asymmetry of synchronization, which is further motivated by the evidence that the interhemispheric asymmetries suspected to be abnormal in SZ originate from the connectivity-related structuring of the cortex. Methods: Thirteen right-handed SZ patients and thirteen matched controls, participated in this study and the multichannel (128) EEGs were recorded for 3-5 minutes at rest. Then, Laplacian EEG (LEEG) were calculated using a 2-D spline. The LEEGs were analysis through calculating the power spectral density using Welch's average periodogram method. Furthermore, using a state-space based multivariate synchronization measure, S-estimator, we analyzed the correlate of the functional cortico-cortical connectivity in SZ patients compared to the controls. The values of S-estimator were obtained at three different special scales: first-order neighbors for each sensor location, second-order neighbors, and the whole hemisphere. The synchronization measures based on LEEG of alpha and beta bands were applied and tuned to various spatial scales including local, intraregional, and long-distance levels. To assess the between-group differences, we used a permutation version of Hotelling's T2 test. For correlation analysis, Spearman Rank Correlation was calculated. Results: Compared to the controls, who had rightward asymmetry at a local level (LEEG power), rightward anterior and leftward posterior asymmetries at an intraregional level (first- and second-order S-estimator), and rightward global asymmetry (hemispheric S-estimator), SZ patients showed generally attenuated asymmetry, the effect being strongest for intraregional synchronization. This deviation in asymmetry across the anterior-to-posterior axis is consistent with the cerebral form of the so-called Yakovlevian or anticlockwise cerebral torque. Moreover, the negative occipital and positive frontal asymmetry values suggest higher regional synchronization among the left occipital and the right frontal locations relative to their symmetrical counterparts. Correlation analysis linked the posterior intraregional and hemispheric abnormalities to the negative SZ symptoms, whereas the asymmetry of LEEG power appeared to be weakly coupled to clinical ratings. The posterior intraregional abnormalities of asymmetry were shown to increase with the duration of the disease. The tentative links between these findings and gross anatomical asymmetries, including the cerebral torque and gyrification pattern in normal subjects and SZ patients, are discussed. Conclusions: Overall, our findings reveal the abnormalities in the synchronization asymmetry in SZ patients and heavy involvement of the right hemisphere in these abnormalities. These results indicate that anomalous asymmetry of cortico-cortical connections in schizophrenia is amenable to electrophysiological analysis.

Sequential estimation of time-varying multipath channel for MIMO-OFDM systems

In this paper, we introduce a pilot-aided multipath channel estimator for Multiple-Input Multiple-Output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) systems. Typical estimation algorithms assume the number of multipath components and delays to be known and constant, while theiramplitudes may vary in time. In this work, we focus on the more realistic assumption that also the number of channel taps is unknown and time-varying. The estimation problem arising from this assumption is solved using Random Set Theory (RST), which is a probability theory of finite sets. Due to the lack of a closed form of the optimal filter, a Rao-Blackwellized Particle Filter (RBPF) implementation of the channel estimator is derived. Simulation results demonstrate the estimator effectiveness.

An evaluation of new parsimony-based versus parametric inference methods in biogeography: a case study using the globally distributed plant family Sapindaceae

Aim  Recently developed parametric methods in historical biogeography allow researchers to integrate temporal and palaeogeographical information into the reconstruction of biogeographical scenarios, thus overcoming a known bias of parsimony-based approaches. Here, we compare a parametric method, dispersal-extinction-cladogenesis (DEC), against a parsimony-based method, dispersal-vicariance analysis (DIVA), which does not incorporate branch lengths but accounts for phylogenetic uncertainty through a Bayesian empirical approach (Bayes-DIVA). We analyse the benefits and limitations of each method using the cosmopolitan plant family Sapindaceae as a case study.Location  World-wide.Methods  Phylogenetic relationships were estimated by Bayesian inference on a large dataset representing generic diversity within Sapindaceae. Lineage divergence times were estimated by penalized likelihood over a sample of trees from the posterior distribution of the phylogeny to account for dating uncertainty in biogeographical reconstructions. We compared biogeographical scenarios between Bayes-DIVA and two different DEC models: one with no geological constraints and another that employed a stratified palaeogeographical model in which dispersal rates were scaled according to area connectivity across four time slices, reflecting the changing continental configuration over the last 110 million years.Results  Despite differences in the underlying biogeographical model, Bayes-DIVA and DEC inferred similar biogeographical scenarios. The main differences were: (1) in the timing of dispersal events - which in Bayes-DIVA sometimes conflicts with palaeogeographical information, and (2) in the lower frequency of terminal dispersal events inferred by DEC. Uncertainty in divergence time estimations influenced both the inference of ancestral ranges and the decisiveness with which an area can be assigned to a node.Main conclusions  By considering lineage divergence times, the DEC method gives more accurate reconstructions that are in agreement with palaeogeographical evidence. In contrast, Bayes-DIVA showed the highest decisiveness in unequivocally reconstructing ancestral ranges, probably reflecting its ability to integrate phylogenetic uncertainty. Care should be taken in defining the palaeogeographical model in DEC because of the possibility of overestimating the frequency of extinction events, or of inferring ancestral ranges that are outside the extant species ranges, owing to dispersal constraints enforced by the model. The wide-spanning spatial and temporal model proposed here could prove useful for testing large-scale biogeographical patterns in plants.

Evaluation of adding a second marker to overcome Staphylococcus aureus spa typing homoplasies.

The utility of sequencing a second highly variable locus in addition to the spa gene (e.g., double-locus sequence typing [DLST]) was investigated to overcome limitations of a Staphylococcus aureus single-locus typing method. Although adding a second locus seemed to increase discriminatory power, it was not sufficient to definitively infer evolutionary relationships within a single multilocus sequence type (ST-5).

Substantial advantage of a combined Bayesian and genotyping approach in testosterone doping tests.

Testosterone abuse is conventionally assessed by the urinary testosterone/epitestosterone (T/E) ratio, levels above 4.0 being considered suspicious. A deletion polymorphism in the gene coding for UGT2B17 is strongly associated with reduced testosterone glucuronide (TG) levels in urine. Many of the individuals devoid of the gene would not reach a T/E ratio of 4.0 after testosterone intake. Future test programs will most likely shift from population based- to individual-based T/E cut-off ratios using Bayesian inference. A longitudinal analysis is dependent on an individual's true negative baseline T/E ratio. The aim was to investigate whether it is possible to increase the sensitivity and specificity of the T/E test by addition of UGT2B17 genotype information in a Bayesian framework. A single intramuscular dose of 500mg testosterone enanthate was given to 55 healthy male volunteers with either two, one or no allele (ins/ins, ins/del or del/del) of the UGT2B17 gene. Urinary excretion of TG and the T/E ratio was measured during 15 days. The Bayesian analysis was conducted to calculate the individual T/E cut-off ratio. When adding the genotype information, the program returned lower individual cut-off ratios in all del/del subjects increasing the sensitivity of the test considerably. It will be difficult, if not impossible, to discriminate between a true negative baseline T/E value and a false negative one without knowledge of the UGT2B17 genotype. UGT2B17 genotype information is crucial, both to decide which initial cut-off ratio to use for an individual, and for increasing the sensitivity of the Bayesian analysis.

Safety Effectiveness of High-Speed Expressway Signals, 2005

The safety benefit of signalizing intersections of high-speed divided expressways is considered. Analyses were conducted on 50 and 55 mph and on 55 mph only intersections, comparing unsignalized and signalized intersections. Results of the 55 mph analysis are included in this report. Matched-pair analysis indicates that generally, signalized intersections have higher crash rate but lower costs per crash. On the other hand, before-and-after analysis (intersections signalized between 1994 and 2001) indicates lower crash rates (~30 percent) and total costs (~10 percent) after signalization. Empirical Bayes (EB) adjusted before-and-after analysis reduces estimates of safety benefit (crash rate) to about 20 percent. The study shows how commonly used analyses can differ in their results, and that there is great variability in the safety performance of individual signalized locations. This variability and the effect of EB adjustment are demonstrated through the use of innovative graphics.

PACIC Instrument: disentangling dimensions using published validation models.

OBJECTIVE: To better understand the structure of the Patient Assessment of Chronic Illness Care (PACIC) instrument. More specifically to test all published validation models, using one single data set and appropriate statistical tools. DESIGN: Validation study using data from cross-sectional survey. PARTICIPANTS: A population-based sample of non-institutionalized adults with diabetes residing in Switzerland (canton of Vaud). MAIN OUTCOME MEASURE: French version of the 20-items PACIC instrument (5-point response scale). We conducted validation analyses using confirmatory factor analysis (CFA). The original five-dimension model and other published models were tested with three types of CFA: based on (i) a Pearson estimator of variance-covariance matrix, (ii) a polychoric correlation matrix and (iii) a likelihood estimation with a multinomial distribution for the manifest variables. All models were assessed using loadings and goodness-of-fit measures. RESULTS: The analytical sample included 406 patients. Mean age was 64.4 years and 59% were men. Median of item responses varied between 1 and 4 (range 1-5), and range of missing values was between 5.7 and 12.3%. Strong floor and ceiling effects were present. Even though loadings of the tested models were relatively high, the only model showing acceptable fit was the 11-item single-dimension model. PACIC was associated with the expected variables of the field. CONCLUSIONS: Our results showed that the model considering 11 items in a single dimension exhibited the best fit for our data. A single score, in complement to the consideration of single-item results, might be used instead of the five dimensions usually described.

Estimating the basic reproductive number from viral sequence data.

Epidemiological processes leave a fingerprint in the pattern of genetic structure of virus populations. Here, we provide a new method to infer epidemiological parameters directly from viral sequence data. The method is based on phylogenetic analysis using a birth-death model (BDM) rather than the commonly used coalescent as the model for the epidemiological transmission of the pathogen. Using the BDM has the advantage that transmission and death rates are estimated independently and therefore enables for the first time the estimation of the basic reproductive number of the pathogen using only sequence data, without further assumptions like the average duration of infection. We apply the method to genetic data of the HIV-1 epidemic in Switzerland.

Financial Development, Financial Fragility, and Growth

This paper studies the apparent contradiction between two strands of the literature on the effects of financial intermediation on economic activity. On the one hand, the empirical growth literature finds a positive effect of financial depth as measured by, for instance, private domestic credit and liquid liabilities (e.g., Levine, Loayza, and Beck 2000). On the other hand, the banking and currency crisis literature finds that monetary aggregates, such as domestic credit, are among the best predictors of crises and their related economic downturns (e.g., Kaminski and Reinhart 1999). The paper accounts for these contrasting effects based on the distinction between the short- and long-run impacts of financial intermediation. Working with a panel of cross-country and time-series observations, the paper estimates an encompassing model of short- and long-run effects using the Pooled Mean Group estimator developed by Pesaran, Shin, and Smith (1999). The conclusion from this analysis is that a positive long-run relationship between financial intermediation and output growth co-exists with a, mostly, negative short-run relationship. The paper further develops an explanation for these contrasting effects by relating them to recent theoretical models, by linking the estimated short-run effects to measures of financial fragility (namely, banking crises and financial volatility), and by jointly analyzing the effects of financial depth and fragility in classic panel growth regressions.