Improved computations for relationship inference using low-coverage sequencing data.

Pedigree inference, for example determining whether two persons are second cousins or unrelated, can be done by comparing their genotypes at a selection of genetic markers. When the data for one or more of the persons is from low-coverage next generation sequencing (lcNGS), currently available computational methods either ignore genetic linkage or do not take advantage of the probabilistic nature of lcNGS data, relying instead on first estimating the genotype. We provide a method and software (see familias.name/lcNGS) bridging the above gap. Simulations indicate how our results are considerably more accurate compared to some previously available alternatives. Our method, utilizing a version of the Lander-Green algorithm, uses a group of symmetries to speed up calculations. This group may be of further interest in other calculations involving linked loci.

Mathematically optimal decisions in forensic age assessment.

Forensic age estimation generally involves considerable amounts of uncertainty. Forensic age indicators such as teeth or skeleton images predict age only approximately, and this is likely to remain true even for future forensic age indicators. Thus, forensic age assessment should aim to make the best possible decisions under uncertainty. In this paper, we apply mathematical theory to make statistically optimal decisions to age assessment. Such an application is fairly straightforward assuming there is a standardized procedure for obtaining age indicator information from individuals, assuming we have data from the application of this procedure to a group of persons with known ages, and assuming the starting point for each individual is a probability distribution describing prior knowledge about the persons age. The main problem is then to obtain such a prior. Our analysis indicates that individual priors rather than a common prior for all persons may be necessary. We suggest that caseworkers, based on individual case information, may select a prior from a menu of priors. We show how information may then be collected over time to gradually increase the robustness of the decision procedure. We also show how replacing individual prior distributions for age with individual prior odds for being above an age limit cannot be recommended as a general method. Our theoretical framework is applied to data where the maturity of the distal femur and the third molar is observed using MRI. As part of this analysis we observe a weak positive conditional correlation between maturity of the two body parts.

Error rates for unvalidated medical age assessment procedures.

During 2014-2015, Sweden received asylum applications from more than 240,000 people, of which more than 40,000 were termed unaccompanied minors. In a large number of cases, claims by asylum seekers of being below 18 years were not trusted by Swedish authorities. To handle the situation, the Swedish national board of forensic medicine (Rättsmedicinalverket, RMV) was assigned by the government to create a centralized system for medical age assessments. RMV introduced a procedure including two biological age indicators; x-ray of the third molars and magnetic resonance imaging of the distal femoral epiphysis. In 2017, a total of 9617 males and 337 females were subjected to this procedure. No validation study for the procedure was however published, and the observed number of cases with different maturity combinations in teeth and femur were unexpected given the claims originally made by RMV. We present a general stochastic model enabling us to study which combinations of age indicator model parameters and age population profiles are consistent with the observed 2017 data for males. We find that, contrary to some RMV claims, maturity of the femur, as observed by RMV, appears on average well before maturity of teeth. According to our estimates, approximately 15% of the tested males were children. These children had an approximate 33% risk of being classified as adults. The corresponding risk for an adult to be misclassified as a child was approximately 7%. We determine uncertainties and ranges of estimates under reasonable perturbations of the prior.

Determining the optimal forensic DNA analysis procedure following investigation of sample quality.

Crime scene traces of various types are routinely sent to forensic laboratories for analysis, generally with the aim of addressing questions about the source of the trace. The laboratory may choose to analyse the samples in different ways depending on the type and quality of the sample, the importance of the case and the cost and performance of the available analysis methods. Theoretically well-founded guidelines for the choice of analysis method are, however, lacking in most situations. In this paper, it is shown how such guidelines can be created using Bayesian decision theory. The theory is applied to forensic DNA analysis, showing how the information from the initial qPCR analysis can be utilized. It is assumed the alternatives for analysis are using a standard short tandem repeat (STR) DNA analysis assay, using the standard assay and a complementary assay, or the analysis may be cancelled following quantification. The decision is based on information about the DNA amount and level of DNA degradation of the forensic sample, as well as case circumstances and the cost for analysis. Semi-continuous electropherogram models are used for simulation of DNA profiles and for computation of likelihood ratios. It is shown how tables and graphs, prepared beforehand, can be used to quickly find the optimal decision in forensic casework.

A general model for likelihood computations of genetic marker data accounting for linkage, linkage disequilibrium, and mutations.

Several applications necessitate an unbiased determination of relatedness, be it in linkage or association studies or in a forensic setting. An appropriate model to compute the joint probability of some genetic data for a set of persons given some hypothesis about the pedigree structure is then required. The increasing number of markers available through high-density SNP microarray typing and NGS technologies intensifies the demand, where using a large number of markers may lead to biased results due to strong dependencies between closely located loci, both within pedigrees (linkage) and in the population (allelic association or linkage disequilibrium (LD)). We present a new general model, based on a Markov chain for inheritance patterns and another Markov chain for founder allele patterns, the latter allowing us to account for LD. We also demonstrate a specific implementation for X chromosomal markers that allows for computation of likelihoods based on hypotheses of alleged relationships and genetic marker data. The algorithm can simultaneously account for linkage, LD, and mutations. We demonstrate its feasibility using simulated examples. The algorithm is implemented in the software FamLinkX, providing a user-friendly GUI for Windows systems (FamLinkX, as well as further usage instructions, is freely available at www.famlink.se ). Our software provides the necessary means to solve cases where no previous implementation exists. In addition, the software has the possibility to perform simulations in order to further study the impact of linkage and LD on computed likelihoods for an arbitrary set of markers.

A phylogenetic comparative method for studying multivariate adaptation.

Phylogenetic comparative methods have been limited in the way they model adaptation. Although some progress has been made, there are still no methods that can fully account for coadaptation between traits. Based on Ornstein-Uhlenbeck (OU) models of adaptive evolution, we present a method, with R implementation, in which multiple traits evolve both in response to each other and, as in previous OU models, to fixed or randomly evolving predictor variables. We present the interpretation of the model parameters in terms of evolutionary and optimal regressions enabling the study of allometric and adaptive relationships between traits. To illustrate the method we reanalyze a data set of antler and body-size evolution in deer (Cervidae).

Application of the Bayesian framework for forensic interpretation to casework involving postmortem interval estimates of decomposed human remains.

We demonstrate how the Bayesian framework for forensic interpretation can be adapted for casework involving postmortem intervals (PMI) utilizing taphonomic data as well as how to overcome some of the limitations of current approaches for estimating and communicating uncertainty. A model is implemented for indoor cases based on partial body scores from three different anatomical regions as correlated functions of accumulated temperature (AT). The multivariate model enables estimation of PMI for human remains also when one or two local body scores are missing or undetermined, e.g. as a result of burns, scars or covered body parts. The model was trained using the expectation maximization algorithm, enabling us to account for uncertainty of PMI and/or ambient temperature in the training data. Alternative approaches reporting the results are presented, including the likelihood curve, likelihood ratios for competing hypotheses and posterior probability distributions and credibility intervals for PMI. The applicability or the approaches in different forensic scenarios is discussed.

Surveillance of animal diseases through implementation of a Bayesian spatio-temporal model: A simulation example with neurological syndromes in horses and West Nile Virus.

A potentially sensitive way to detect disease outbreaks is syndromic surveillance, i.e. monitoring the number of syndromes reported in the population of interest, comparing it to the baseline rate, and drawing conclusions about outbreaks using statistical methods. A decision maker may use the results to take disease control actions or to initiate enhanced epidemiological investigations. In addition to the total count of syndromes there are often additional pieces of information to consider when assessing the probability of an outbreak. This includes clustering of syndromes in space and time as well as historical data on the occurrence of syndromes, seasonality of the disease, etc. In this paper, we show how Bayesian theory for syndromic surveillance applies to the occurrence of neurological syndromes in horses in France. Neurological syndromes in horses may be connected e.g. to West Nile Virus (WNV), a zoonotic disease of growing concern for public health in Europe. A Bayesian method for spatio-temporal cluster detection of syndromes and for determining the probability of an outbreak is presented. It is shown how surveillance can be performed simultaneously for a specific class of diseases (WNV or diseases similar to WNV in terms of the information available to the system) and a non-specific class of diseases (not similar to WNV in terms of the information available to the system). We also discuss some new extensions to the spatio-temporal models and the computational algorithms involved. It is shown step-by-step how data from historical WNV outbreaks and surveillance data for neurological syndromes can be used for model construction. The model is implemented using a Gibbs sampling procedure, and its sensitivity and specificity is evaluated. Finally, it is illustrated how predictive modelling of syndromes can be useful for decision making in animal health surveillance.

Empirical Bayes models for multiple probe type microarrays at the probe level.

BACKGROUND: When analyzing microarray data a primary objective is often to find differentially expressed genes. With empirical Bayes and penalized t-tests the sample variances are adjusted towards a global estimate, producing more stable results compared to ordinary t-tests. However, for Affymetrix type data a clear dependency between variability and intensity-level generally exists, even for logged intensities, most clearly for data at the probe level but also for probe-set summarizes such as the MAS5 expression index. As a consequence, adjustment towards a global estimate results in an intensity-level dependent false positive rate. RESULTS: We propose two new methods for finding differentially expressed genes, Probe level Locally moderated Weighted median-t (PLW) and Locally Moderated Weighted-t (LMW). Both methods use an empirical Bayes model taking the dependency between variability and intensity-level into account. A global covariance matrix is also used allowing for differing variances between arrays as well as array-to-array correlations. PLW is specially designed for Affymetrix type arrays (or other multiple-probe arrays). Instead of making inference on probe-set summaries, comparisons are made separately for each perfect-match probe and are then summarized into one score for the probe-set. CONCLUSION: The proposed methods are compared to 14 existing methods using five spike-in data sets. For RMA and GCRMA processed data, PLW has the most accurate ranking of regulated genes in four out of the five data sets, and LMW consistently performs better than all examined moderated t-tests when used on RMA, GCRMA, and MAS5 expression indexes.

Transcriptome analysis of a respiratory Saccharomyces cerevisiae strain suggests the expression of its phenotype is glucose insensitive and predominantly controlled by Hap4, Cat8 and Mig1.

BACKGROUND: We previously described the first respiratory Saccharomyces cerevisiae strain, KOY.TM6*P, by integrating the gene encoding a chimeric hexose transporter, Tm6*, into the genome of an hxt null yeast. Subsequently we transferred this respiratory phenotype in the presence of up to 50 g/L glucose to a yeast strain, V5 hxt1-7Delta, in which only HXT1-7 had been deleted. In this study, we compared the transcriptome of the resultant strain, V5.TM6*P, with that of its wild-type parent, V5, at different glucose concentrations. RESULTS: cDNA array analyses revealed that alterations in gene expression that occur when transitioning from a respiro-fermentative (V5) to a respiratory (V5.TM6*P) strain, are very similar to those in cells undergoing a diauxic shift. We also undertook an analysis of transcription factor binding sites in our dataset by examining previously-published biological data for Hap4 (in complex with Hap2, 3, 5), Cat8 and Mig1, and used this in combination with verified binding consensus sequences to identify genes likely to be regulated by one or more of these. Of the induced genes in our dataset, 77% had binding sites for the Hap complex, with 72% having at least two. In addition, 13% were found to have a binding site for Cat8 and 21% had a binding site for Mig1. Unexpectedly, both the up- and down-regulation of many of the genes in our dataset had a clear glucose dependence in the parent V5 strain that was not present in V5.TM6*P. This indicates that the relief of glucose repression is already operable at much higher glucose concentrations than is widely accepted and suggests that glucose sensing might occur inside the cell. CONCLUSION: Our dataset gives a remarkably complete view of the involvement of genes in the TCA cycle, glyoxylate cycle and respiratory chain in the expression of the phenotype of V5.TM6*P. Furthermore, 88% of the transcriptional response of the induced genes in our dataset can be related to the potential activities of just three proteins: Hap4, Cat8 and Mig1. Overall, our data support genetic remodelling in V5.TM6*P consistent with a respiratory metabolism which is insensitive to external glucose concentrations.

HeliCis: a DNA motif discovery tool for colocalized motif pairs with periodic spacing.

BACKGROUND: Correct temporal and spatial gene expression during metazoan development relies on combinatorial interactions between different transcription factors. As a consequence, cis-regulatory elements often colocalize in clusters termed cis-regulatory modules. These may have requirements on organizational features such as spacing, order and helical phasing (periodic spacing) between binding sites. Due to the turning of the DNA helix, a small modification of the distance between a pair of sites may sometimes drastically disrupt function, while insertion of a full helical turn of DNA (10-11 bp) between cis elements may cause functionality to be restored. Recently, de novo motif discovery methods which incorporate organizational properties such as colocalization and order preferences have been developed, but there are no tools which incorporate periodic spacing into the model. RESULTS: We have developed a web based motif discovery tool, HeliCis, which features a flexible model which allows de novo detection of motifs with periodic spacing. Depending on the parameter settings it may also be used for discovering colocalized motifs without periodicity or motifs separated by a fixed gap of known or unknown length. We show on simulated data that it can efficiently capture the synergistic effects of colocalization and periodic spacing to improve detection of weak DNA motifs. It provides a simple to use web interface which interactively visualizes the current settings and thereby makes it easy to understand the parameters and the model structure. CONCLUSION: HeliCis provides simple and efficient de novo discovery of colocalized DNA motif pairs, with or without periodic spacing. Our evaluations show that it can detect weak periodic patterns which are not easily discovered using a sequential approach, i.e. first finding the binding sites and second analyzing the properties of their pairwise distances.

Improved covariance matrix estimators for weighted analysis of microarray data.

Empirical Bayes models have been shown to be powerful tools for identifying differentially expressed genes from gene expression microarray data. An example is the WAME model, where a global covariance matrix accounts for array-to-array correlations as well as differing variances between arrays. However, the existing method for estimating the covariance matrix is very computationally intensive and the estimator is biased when data contains many regulated genes. In this paper, two new methods for estimating the covariance matrix are proposed. The first method is a direct application of the EM algorithm for fitting the multivariate t-distribution of the WAME model. In the second method, a prior distribution for the log fold-change is added to the WAME model, and a discrete approximation is used for this prior. Both methods are evaluated using simulated and real data. The first method shows equal performance compared to the existing method in terms of bias and variability, but is superior in terms of computer time. For large data sets (>15 arrays), the second method also shows superior computer run time. Moreover, for simulated data with regulated genes the second method greatly reduces the bias. With the proposed methods it is possible to apply the WAME model to large data sets with reasonable computer run times. The second method shows a small bias for simulated data, but appears to have a larger bias for real data with many regulated genes.

Detection probability models for bacteria, and how to obtain them from heterogeneous spiking data. An application to Bacillus anthracis.

Efficient and correct evaluation of sampling results with respect to hypotheses about the concentration or distribution of bacteria generally requires knowledge about the performance of the detection method. To assess the sensitivity of the detection method an experiment is usually performed where the target matrix is spiked (i.e. artificially contaminated) with different concentrations of the bacteria, followed by analyses of the samples using the pre-enrichment method and the analytical detection method of interest. For safety reasons or because of economic or time limits it is not always possible to perform exactly such an experiment, with the desired number of samples. In this paper, we show how heterogeneous data from diverse sources may be combined within a single model to obtain not only estimates of detection probabilities, but also, crucially, uncertainty estimates. We indicate how such results can then be used to obtain optimal conclusions about presence of bacteria, and illustrate how strongly the sampling results speak in favour of or against contamination. In our example, we consider the case when B. cereus is used as surrogate for B. anthracis, for safety reasons. The statistical modelling of the detection probabilities and of the growth characteristics of the bacteria types is based on data from four experiments where different matrices of food were spiked with B. anthracis or B. cereus and analysed using plate counts and qPCR. We show how flexible and complex Bayesian models, together with inference tools such as OpenBUGS, can be used to merge information about detection probability curves. Two different modelling approaches, differing in whether the pre-enrichment step and the PCR detection step are modelled separately or together, are applied. The relative importance on the detection curves for various existing data sets are evaluated and illustrated.

Stationary mutation models.

Probability calculations for relationship inference based on DNA tests are often performed with computer packages such as Familias. When mutations are assumed to be a possibility, one may notice a curious and problematic effect of including untested parents: results tend to change slightly. In this paper, we trace this effect back to fundamental model-formulating issues which can only be resolved by using stationary mutation models. We present several methods for obtaining such stationary mutation matrices from original mutation matrices, and evaluate essential properties of these methods. Our conclusion is that typically, stationary mutation models can be obtained, but for many types of markers, it may be impossible to combine specific biologically reasonable requirements for a mutation matrix with the requirement of stationarity.

Design of improved membrane protein production experiments: quantitation of the host response.

Eukaryotic membrane proteins cannot be produced in a reliable manner for structural analysis. Consequently, researchers still rely on trial-and-error approaches, which most often yield insufficient amounts. This means that membrane protein production is recognized by biologists as the primary bottleneck in contemporary structural genomics programs. Here, we describe a study to examine the reasons for successes and failures in recombinant membrane protein production in yeast, at the level of the host cell, by systematically quantifying cultures in high-performance bioreactors under tightly-defined growth regimes. Our data show that the most rapid growth conditions of those chosen are not the optimal production conditions. Furthermore, the growth phase at which the cells are harvested is critical: We show that it is crucial to grow cells under tightly-controlled conditions and to harvest them prior to glucose exhaustion, just before the diauxic shift. The differences in membrane protein yields that we observe under different culture conditions are not reflected in corresponding changes in mRNA levels of FPS1, but rather can be related to the differential expression of genes involved in membrane protein secretion and yeast cellular physiology.

Predictive screening for regulators of conserved functional gene modules (gene batteries) in mammals.

BACKGROUND: The expression of gene batteries, genomic units of functionally linked genes which are activated by similar sets of cis- and trans-acting regulators, has been proposed as a major determinant of cell specialization in metazoans. We developed a predictive procedure to screen the mouse and human genomes and transcriptomes for cases of gene-battery-like regulation. RESULTS: In a screen that covered approximately 40 percent of all annotated protein-coding genes, we identified 21 co-expressed gene clusters with statistically supported sharing of cis-regulatory sequence elements. 66 predicted cases of over-represented transcription factor binding motifs were validated against the literature and fell into three categories: (i) previously described cases of gene battery-like regulation, (ii) previously unreported cases of gene battery-like regulation with some support in a limited number of genes, and (iii) predicted cases that currently lack experimental support. The novel predictions include for example Sox 17 and RFX transcription factor binding sites that were detected in approximately 10% of all testis specific genes, and HNF-1 and 4 binding sites that were detected in approximately 30% of all kidney specific genes respectively. The results are publicly available at http://www.wlab.gu.se/lindahl/genebatteries. CONCLUSION: 21 co-expressed gene clusters were enriched for a total of 66 shared cis-regulatory sequence elements. A majority of these predictions represent novel cases of potential co-regulation of functionally coupled proteins. Critical technical parameters were evaluated, and the results and the methods provide a valuable resource for future experimental design.

Enhanced low-template DNA analysis conditions and investigation of allele dropout patterns.

Forensic DNA analysis applying PCR enables profiling of minute biological samples. Enhanced analysis conditions can be applied to further push the limit of detection, coming with the risk of visualising artefacts and allele imbalances. We have evaluated the consecutive increase of PCR cycles from 30 to 35 to investigate the limitations of low-template (LT) DNA analysis, applying the short tandem repeat (STR) analysis kit PowerPlex ESX 16. Mock crime scene DNA extracts of four different quantities (from around 8-84 pg) were tested. All PCR products were analysed using 5, 10 and 20 capillary electrophoresis (CE) injection seconds. Bayesian models describing allele dropout patterns, allele peak heights and heterozygote balance were developed to assess the overall improvements in EPG quality with altered PCR/CE settings. The models were also used to evaluate the impact of amplicon length, STR marker and fluorescent label on the risk for allele dropout. The allele dropout probability decreased for each PCR cycle increment from 30 to 33 PCR cycles. Irrespective of DNA amount, the dropout probability was not affected by further increasing the number of PCR cycles. For the 42 and 84 pg samples, mainly complete DNA profiles were generated applying 32 PCR cycles. For the 8 and 17 pg samples, the allele dropouts decreased from 100% using 30 cycles to about 75% and 20%, respectively. The results for 33, 34 and 35 PCR cycles indicated that heterozygote balance and stutter ratio were mainly affected by DNA amount, and not directly by PCR cycle number and CE injection settings. We found 32 and 33 PCR cycles with 10 CE injection seconds to be optimal, as 34 and 35 PCR cycles did not improve allele detection and also included CE saturation problems. We find allele dropout probability differences between several STR markers. Markers labelled with the fluorescent dyes CXR-ET (red in electropherogram) and TMR-ET (shown as black) generally have higher dropout risks compared with those labelled with JOE (green) and fluorescein (blue). Overall, the marker D10S1248 has the lowest allele dropout probability and D8S1179 the highest. The marker effect is mainly pronounced for 30-32 PCR cycles. Such effects would not be expected if the amplification efficiencies were identical for all markers. Understanding allele dropout risks and the variability in peak heights and balances is important for correct interpretation of forensic DNA profiles.

A clustering algorithm using DNA marker information for sub-pedigree reconstruction.

In a mass disaster scenario in which many people are dead, it may be that small family groups are among the dead, and investigators may need to identify such groups, e.g., to return bodies to living relatives for burial. We consider the problem of identifying small groups of closely related people within a large group of people through the use of DNA marker information. We propose a likelihood-ratio-based distance measure of the relatedness between pairs of individuals and use an estimate of this measure as a means of clustering related people into groups. We show the effectiveness of our approach on real examples and through simulations, which suggest that the method is quite reliable for identifying very close relationships. We discuss the use of our clustering algorithm in a two-stage pedigree reconstruction procedure and suggest directions in which the analysis could be extended. Applications include the identification of family groups among bodies found in mass graves and identification of family groups in animal populations.

Choosing supplementary markers in forensic casework.

The vast majority of human familial identifications based on DNA end up with a well founded conclusion, normally using a standard set of genetic short tandem repeat (STR) loci. There are, however, a proportion of cases that show ambiguous results. For such occasions a number of different supplementary markers could be typed in order to gain further information. There are numerous markers available for such supplementary DNA typing, including STRs, deletion and insertion polymorphisms (DIPs), and single nucleotide polymorphisms (SNPs). The purpose of this work was to describe a precise method for decision making, aiming to aid the comparison of different sets of markers for different case scenarios in order to find the most efficient set for routine casework. Comparisons are based on a particular function relating the expected additional value of information from new data to the amount of information already obtained from initial data. The function can be computed approximately by approximating likelihood-based error rates using simulation. In this paper we focused on paternity investigations, more specifically the use of supplementary markers in cases where a smaller number of genetic inconsistencies make the matter inconclusive. We applied the method to a comparison of three different kits: Investigator HDplex (STRs), Investigator DIPplex (DIPs), and the SNPforID-plex (SNPs) to study their efficiencies in gaining information in different case scenarios involving various alternative relationships between the tested man and the tested child. We show that the Investigator HDplex was the most efficient set of supplementary markers for the standard paternity case. However, for paternity cases with a close relative being the alternative father, the Investigator HDplex and the SNPforID-plex showed similar patterns in their ability to deliver a well-founded conclusion. The Investigator DIPplex was the least efficient set.

Using object oriented bayesian networks to model linkage, linkage disequilibrium and mutations between STR markers.

In a number of applications there is a need to determine the most likely pedigree for a group of persons based on genetic markers. Adequate models are needed to reach this goal. The markers used to perform the statistical calculations can be linked and there may also be linkage disequilibrium (LD) in the population. The purpose of this paper is to present a graphical Bayesian Network framework to deal with such data. Potential LD is normally ignored and it is important to verify that the resulting calculations are not biased. Even if linkage does not influence results for regular paternity cases, it may have substantial impact on likelihood ratios involving other, more extended pedigrees. Models for LD influence likelihoods for all pedigrees to some degree and an initial estimate of the impact of ignoring LD and/or linkage is desirable, going beyond mere rules of thumb based on marker distance. Furthermore, we show how one can readily include a mutation model in the Bayesian Network; extending other programs or formulas to include such models may require considerable amounts of work and will in many case not be practical. As an example, we consider the two STR markers vWa and D12S391. We estimate probabilities for population haplotypes to account for LD using a method based on data from trios, while an estimate for the degree of linkage is taken from the literature. The results show that accounting for haplotype frequencies is unnecessary in most cases for this specific pair of markers. When doing calculations on regular paternity cases, the markers can be considered statistically independent. In more complex cases of disputed relatedness, for instance cases involving siblings or so-called deficient cases, or when small differences in the LR matter, independence should not be assumed. (The networks are freely available at http://arken.umb.no/~dakl/BayesianNetworks.).