Unbiased and efficient log-likelihood estimation with inverse binomial sampling

The fate of scientific hypotheses often relies on the ability of a computational model to explain the data, quantified in modern statistical approaches by the likelihood function. The log-likelihood is the key element for parameter estimation and model evaluation. However, the log-likelihood of complex models in fields such as computational biology and neuroscience is often intractable to compute analytically or numerically. In those cases, researchers can often only estimate the log-likelihood by comparing observed data with synthetic observations generated by model simulations. Standard techniques to approximate the likelihood via simulation either use summary statistics of the data or are at risk of producing substantial biases in the estimate. Here, we explore another method, inverse binomial sampling (IBS), which can estimate the log-likelihood of an entire data set efficiently and without bias. For each observation, IBS draws samples from the simulator model until one matches the observation. The log-likelihood estimate is then a function of the number of samples drawn. The variance of this estimator is uniformly bounded, achieves the minimum variance for an unbiased estimator, and we can compute calibrated estimates of the variance. We provide theoretical arguments in favor of IBS and an empirical assessment of the method for maximum-likelihood estimation with simulation-based models. As case studies, we take three model-fitting problems of increasing complexity from computational and cognitive neuroscience. In all problems, IBS generally produces lower error in the estimated parameters and maximum log-likelihood values than alternative sampling methods with the same average number of samples. Our results demonstrate the potential of IBS as a practical, robust, and easy to implement method for log-likelihood evaluation when exact techniques are not available.     

Convenient Method of Micrometer-Scale Excitation of Propagating Surface Plasmons by a Focused Laser Beam

Plasmonics - Surface plasmons (SPs) excited on microscopic areas are desired in a variety of applications, including SP microscopy and (bio)sensing. We present an optical method of excitation of...     

The Causes and Evolutionary Consequences of Mixed Singing in Two Hybridizing Songbird Species (Luscinia spp.)

Bird song plays an important role in the establishment and maintenance of prezygotic reproductive barriers. When two closely related species come into secondary contact, song convergence caused by acquisition of heterospecific songs into the birds’ repertoires is often observed. The proximate mechanisms responsible for such mixed singing, and its effect on the speciation process, are poorly understood. We used a combination of genetic and bioacoustic analyses to test whether mixed singing observed in the secondary contact zone of two passerine birds, the Thrush Nightingale (Luscinia luscinia) and the Common Nightingale (L. megarhynchos), is caused by introgressive hybridization. We analysed song recordings of both species from allopatric and sympatric populations together with genotype data from one mitochondrial and seven nuclear loci. Semi-automated comparisons of our recordings with an extensive catalogue of Common Nightingale song types confirmed that most of the analysed sympatric Thrush Nightingale males were ‘mixed singers’ that use heterospecific song types in their repertoires. None of these ‘mixed singers’ possessed any alleles introgressed from the Common Nightingale, suggesting that they were not backcross hybrids. We also analysed songs of five individuals with intermediate phenotype, which were identified as F₁ hybrids between the Thrush Nightingale female and the Common Nightingale male by genetic analysis. Songs of three of these hybrids corresponded to the paternal species (Common Nightingale) but the remaining two sung a mixed song. Our results suggest that although hybridization might increase the tendency for learning songs from both parental species, interspecific cultural transmission is the major proximate mechanism explaining the occurrence of mixed singers among the sympatric Thrush Nightingales. We also provide evidence that mixed singing does not substantially increase the rate of interspecific hybridization and discuss the possible adaptive value of this phenomenon in nightingales.     

The peripheral chimerism of bone marrow–derived stem cells after transplantation: regeneration of gastrointestinal tissues in lethally irradiated mice

Bone marrow–derived cells represent a heterogeneous cell population containing haematopoietic stem and progenitor cells. These cells have been identified as potential candidates for use in cell therapy for the regeneration of damaged tissues caused by trauma, degenerative diseases, ischaemia and inflammation or cancer treatment. In our study, we examined a model using whole-body irradiation and the transplantation of bone marrow (BM) or haematopoietic stem cells (HSCs) to study the repair of haematopoiesis, extramedullary haematopoiesis and the migration of green fluorescent protein (GFP⁺) transplanted cells into non-haematopoietic tissues. We investigated the repair of damage to the BM, peripheral blood, spleen and thymus and assessed the ability of this treatment to induce the entry of BM cells or GFP⁺lin⁻Sca-1⁺ cells into non-haematopoietic tissues. The transplantation of BM cells or GFP⁺lin⁻Sca-1⁺ cells from GFP transgenic mice successfully repopulated haematopoiesis and the haematopoietic niche in haematopoietic tissues, specifically the BM, spleen and thymus. The transplanted GFP⁺ cells also entered the gastrointestinal tract (GIT) following whole-body irradiation. Our results demonstrate that whole-body irradiation does not significantly alter the integrity of tissues such as those in the small intestine and liver. Whole-body irradiation also induced myeloablation and chimerism in tissues, and induced the entry of transplanted cells into the small intestine and liver. This result demonstrates that grafted BM cells or GFP⁺lin⁻Sca-1⁺ cells are not transient in the GIT. Thus, these transplanted cells could be used for the long-term treatment of various pathologies or as a one-time treatment option if myeloablation-induced chimerism alone is not sufficient to induce the entry of transplanted cells into non-haematopoietic tissues.     

Study of strategies for selecting quantitative trait locus mapping procedures by computer simulation

Along with the development and integration of molecular genetics and quantitative genetics, many quantitative trait locus (QTL) mapping studies have been conducted using different mapping populations in various crop species. Existing QTLs can be used for marker-assisted breeding and map-based cloning, whereas the false-positive QTLs are no use. The purpose of this study is to evaluate the suitability of different mapping procedures for data from different genetic models. In this study, four types of recombinant inbred lines (RILs) with different genetic models, viz. additive QTLs (Model I), additive and epistatic QTLs (Model II), additive QTLs and QTL × environment interaction (Model III), additive, epistatic QTLs and QTL × environment interaction (Model IV), were simulated by computer. Six types of QTL mapping procedures, viz. CIM, MIMF, MIMR, ICIM, MQM and NWIM, on four kinds of QTL mapping software, viz. WinQTL Cartographer Version 2.5, IciMapping Version 2.0, MapQTL Version 5.0 and QTLnetwork Version 2.0, were used for screening QTLs of the simulated RILs. The results showed that different mapping procedures have different suitability for different genetic models. CIM and MQM can only screen Model I data. MIMR, MIMF and ICIM can only screen Model I and Model II data. NWIM can screen all four models’ data. It can be concluded that different genetic models’ data have different most suitable mapping procedures. In practical experiments where the genetic model of the data is unknown, a multiple model mapping strategy should be used, that is a full model scanning with complex model procedure followed by verification with other procedures corresponding to the scanning results.     

Reproductive competition between females in the matrilineal Mosuo of southwestern China

The matrilineal Mosuo of southwestern China live in communal households where brothers and sisters of three generations live together (duolocal residence), and men visit their wives, who reside elsewhere, only at night in ‘visiting’ marriages. Here we show that these communally breeding sisters are in reproductive conflict, in the sense that they share the resources needed to reproduce. We analyse determinants of reproductive success in females and males, and show that co-resident female kin are in competition; the more female kin reside in the household, the more reproductive success is reduced. Male reproductive success, however, is not determined by the kin in his natal household; duolocal males are not in reproductive conflict with their siblings. Competition with female cousins can be worse than that between sisters. We also find that female work on the farm (which is the main communal resource) is not equal. We use a ‘tug-of-war’ model of reproductive skew generated by incomplete control, to model the patterns of effort put into competition between sisters and cousins. The model predicts that more dominant (older) sisters will put less effort into reproductive conflict than will less dominant (younger) sisters; but younger sisters will also have lower reproductive success because they are less efficient at gaining access to the shared resource. Both predictions are consistent with our data. Younger sisters work less in the fields than do older sisters, which may represent a form of conflict or may be because their average relatedness to the household is lower than that of their more fertile older sisters.