Accounting for control mislabeling in case-control biomarker studies

In biomarker discovery studies, uncertainty associated with case and control labels is often overlooked. By omitting to take into account label uncertainty, model parameters and the predictive risk can become biased, sometimes severely. The most common situation is when the control set contains an unknown number of undiagnosed, or future, cases. This has a marked impact in situations where the model needs to be well-calibrated, e.g., when the prediction performance of a biomarker panel is evaluated. Failing to account for class label uncertainty may lead to underestimation of classification performance and bias in parameter estimates. This can further impact on meta-analysis for combining evidence from multiple studies. Using a simulation study, we outline how conventional statistical models can be modified to address class label uncertainty leading to well-calibrated prediction performance estimates and reduced bias in meta-analysis. We focus on the problem of mislabeled control subjects in case-control studies, i.e., when some of the control subjects are undiagnosed cases, although the procedures we report are generic. The uncertainty in control status is a particular situation common in biomarker discovery studies in the context of genomic and molecular epidemiology, where control subjects are commonly sampled from the general population with an established expected disease incidence rate.     

Genetic determinants of serum testosterone concentrations in men

Testosterone concentrations in men are associated with cardiovascular morbidity, osteoporosis, and mortality and are affected by age, smoking, and obesity. Because of serum testosterone''s high heritability, we performed a meta-analysis of genome-wide association data in 8,938 men from seven cohorts and followed up the genome-wide significant findings in one in silico (n = 871) and two de novo replication cohorts (n = 4,620) to identify genetic loci significantly associated with serum testosterone concentration in men. All these loci were also associated with low serum testosterone concentration defined as <300 ng/dl. Two single-nucleotide polymorphisms at the sex hormone-binding globulin (SHBG) locus (17p13-p12) were identified as independently associated with serum testosterone concentration (rs12150660, p = 1.2×10⁻⁴¹ and rs6258, p = 2.3×10⁻²²). Subjects with ≥3 risk alleles of these variants had 6.5-fold higher risk of having low serum testosterone than subjects with no risk allele. The rs5934505 polymorphism near FAM9B on the X chromosome was also associated with testosterone concentrations (p = 5.6×10⁻¹⁶). The rs6258 polymorphism in exon 4 of SHBG affected SHBG''s affinity for binding testosterone and the measured free testosterone fraction (p<0.01). Genetic variants in the SHBG locus and on the X chromosome are associated with a substantial variation in testosterone concentrations and increased risk of low testosterone. rs6258 is the first reported SHBG polymorphism, which affects testosterone binding to SHBG and the free testosterone fraction and could therefore influence the calculation of free testosterone using law-of-mass-action equation.     

The GD1a glycan is a cellular receptor for adenoviruses causing epidemic keratoconjunctivitis

Adenovirus type 37 (Ad37) is a leading cause of epidemic keratoconjunctivitis (EKC), a severe and highly contagious ocular disease. Whereas most other adenoviruses infect cells by engaging CD46 or the coxsackie and adenovirus receptor (CAR), Ad37 binds previously unknown sialic acid-containing cell surface molecules. By glycan array screening, we show here that the receptor-recognizing knob domain of the Ad37 fiber protein specifically binds a branched hexasaccharide that is present in the GD1a ganglioside and that features two terminal sialic acids. Soluble GD1a glycan and GD1a-binding antibodies efficiently prevented Ad37 virions from binding and infecting corneal cells. Unexpectedly, the receptor is constituted by one or more glycoproteins containing the GD1a glycan motif rather than the ganglioside itself, as shown by binding, infection and flow cytometry experiments. Molecular modeling, nuclear magnetic resonance and X-ray crystallography reveal that the two terminal sialic acids dock into two of three previously established sialic acid-binding sites in the trimeric Ad37 knob. Surface plasmon resonance analysis shows that the knob-GD1a glycan interaction has high affinity. Our findings therefore form a basis for the design and development of sialic acid-containing antiviral drugs for topical treatment of EKC.     

Monitoring spatiotemporal variation in abundance and dispersal by a pheromone-kairomone system in the threatened saproxylic beetles <Emphasis Type="Italic">Osmoderma eremita</Emphasis> and <Emphasis Type="Italic">Elater ferrugineus</Emphasis>

Pheromone monitoring could provide unique spatial and temporal information about rare and threatened insect species for conservation purposes. Pheromone traps may be especially valuable in detecting fluctuations and declines in vulnerable species, if trap catch can be related to population density. We exploited the pheromone-kairomone system of the hermit beetle Osmoderma eremita and its predator, the red click beetle Elater ferrugineus, to monitor variation in their flight activity during 7 years (2001–2002, 2005–2009). Spatial and temporal flight activity of O. eremita and E. ferrugineus were highly correlated over individual trap replicates and trap days. Yearly flight activities of both species were correlated within two core sites included during all years of the study, and positively affected by temperature both within and between years. Flight activity could not be directly translated to variation in abundance in either species, however. Dispersal rate was likely the main factor explaining flight activity in O. eremita. Normalizing the flight activity of E. ferrugineus against that of O. eremita did not eliminate most of the variability, however, suggesting that flight activity of E. ferrugineus was governed both by dispersal rate and by population density. Higher fluctuations in population density of E. ferrugineus likely render it more vulnerable to local extinction, which may explain the greater rarity of this species. We suggest that preserving large assemblies of suitable hollow trees would be absolutely essential for buffering against stochastic population fluctuations and securing the long-term persistence of E. ferrugineus.     

Five years of FISH-BOL: brief status report

The Fish Barcode of Life Initiative (FISH-BOL) is a concerted global research project launched in 2005, with the goal to collect and assemble standardized DNA barcode sequences and associated voucher provenance data in a curated reference sequence library to aid the molecular identification of all fish species. This article is a detailed progress report (July 2010) on the number of fish species that have been assigned a DNA barcode. Of the approximately 31,000 currently known fish species, 25% have been processed successfully, with at least one species from 89% of all families barcoded; in this report we give a progress overview by taxonomy and geographic region. Using standard analytical protocols, differences in the barcoding completion rate between orders and families are observed, suggesting a potential PCR amplification bias. Overall, between 3 and 9% of the species analyzed failed to yield a "BARCODE compliant" sequence, depending upon how the data are filtered. When species with only a single representative specimen are included, the failure rate was 9%. This might derive from several sources such as mismatched primers and degraded DNA templates. In an attempt to account for the latter, when the analysis is restricted to species with at least two specimens examined, the observed failure rate is significantly lower (3%), suggesting that template quality is a source of concern for FISH-BOL. We, therefore, conclude that using a standard protocol with several specimens per species and PCR primer cocktails is an efficient and successful approach because failures were evenly distributed among orders and families. Only six orders with low species numbers (Pristiformes, Torpediniformes, Albuliformes, Batrachoidiformes, Gobiesociformes, and Petromyzontiformes) showed failure rates between 10 and 33%. Besides outlining an overarching approach for FISH-BOL data curation, the goal of the present article is to give guidance in directing sampling campaigns toward neglected or underrepresented families in order to complete the FISH-BOL campaign most efficiently.     

HIV-1 molecular epidemiology in Guinea-Bissau, West Africa: origin, demography and migrations

The HIV-1 epidemic in West Africa has been dominated by subtype A and the recombinant form CRF02_AG. Little is known about the origins and the evolutionary history of HIV-1 in this region. We employed Maximum likelihood and Bayesian methods in combination with temporal and spatial information to reconstruct the HIV-1 subtype distribution, demographic history and migration patterns over time in Guinea-Bissau, West Africa. We found that CRF02_AG and subsubtype A3 were the dominant forms of HIV-1 in Guinea-Bissau and that they were introduced into the country on at least six different occasions between 1976 and 1981. These estimates also corresponded well with the first reported HIV-1 cases in Guinea-Bissau. Migration analyses suggested that (1) the HIV-1 epidemic started in the capital Bissau and then dispersed into more rural areas, and (2) the epidemic in Guinea-Bissau was connected to both Cameroon and Mali. This is the first study that describes the HIV-1 molecular epidemiology in a West African country by combining the results of subtype distribution with analyses of epidemic origin and epidemiological linkage between locations. The multiple introductions of HIV-1 into Guinea-Bissau, during a short time-period of five years, coincided with and were likely influenced by the major immigration wave into the country that followed the end of the independence war (1963-1974).     

Characterizing associations and SNP-environment interactions for GWAS-identified prostate cancer risk markers--results from BPC3

Genome-wide association studies (GWAS) have identified multiple single nucleotide polymorphisms (SNPs) associated with prostate cancer risk. However, whether these associations can be consistently replicated, vary with disease aggressiveness (tumor stage and grade) and/or interact with non-genetic potential risk factors or other SNPs is unknown. We therefore genotyped 39 SNPs from regions identified by several prostate cancer GWAS in 10,501 prostate cancer cases and 10,831 controls from the NCI Breast and Prostate Cancer Cohort Consortium (BPC3). We replicated 36 out of 39 SNPs (P-values ranging from 0.01 to 10⁻²⁸). Two SNPs located near KLK3 associated with PSA levels showed differential association with Gleason grade (rs2735839, P = 0.0001 and rs266849, P = 0.0004; case-only test), where the alleles associated with decreasing PSA levels were inversely associated with low-grade (as defined by Gleason grade <8) tumors but positively associated with high-grade tumors. No other SNP showed differential associations according to disease stage or grade. We observed no effect modification by SNP for association with age at diagnosis, family history of prostate cancer, diabetes, BMI, height, smoking or alcohol intake. Moreover, we found no evidence of pair-wise SNP-SNP interactions. While these SNPs represent new independent risk factors for prostate cancer, we saw little evidence for effect modification by other SNPs or by the environmental factors examined.     

The membrane bound LRR lipoprotein Slr, and the cell wall-anchored M1 protein from Streptococcus pyogenes both interact with type I collagen

Streptococcus pyogenes is an important human pathogen and surface structures allow it to adhere to, colonize and invade the human host. Proteins containing leucine rich repeats (LRR) have been identified in mammals, viruses, archaea and several bacterial species. The LRRs are often involved in protein-protein interaction, are typically 20-30 amino acids long and the defining feature of the LRR motif is an 11-residue sequence LxxLxLxxNxL (x being any amino acid). The streptococcal leucine rich (Slr) protein is a hypothetical lipoprotein that has been shown to be involved in virulence, but at present no ligands for Slr have been identified. We could establish that Slr is a membrane attached horseshoe shaped lipoprotein by homology modeling, signal peptidase II inhibition, electron microscopy (of bacteria and purified protein) and immunoblotting. Based on our previous knowledge of LRR proteins we hypothesized that Slr could mediate binding to collagen. We could show by surface plasmon resonance that recombinant Slr and purified M1 protein bind with high affinity to collagen I. Isogenic slr mutant strain (MB1) and emm1 mutant strain (MC25) had reduced binding to collagen type I as shown by slot blot and surface plasmon resonance. Electron microscopy using gold labeled Slr showed multiple binding sites to collagen I, both to the monomeric and the fibrillar structure, and most binding occurred in the overlap region of the collagen I fibril. In conclusion, we show that Slr is an abundant membrane bound lipoprotein that is co-expressed on the surface with M1, and that both these proteins are involved in recruiting collagen type I to the bacterial surface. This underlines the importance of S. pyogenes interaction with extracellular matrix molecules, especially since both Slr and M1 have been shown to be virulence factors.     

Performance of ultra-deep pyrosequencing in analysis of HIV-1 pol gene variation

Introduction

Ultra-deep pyrosequencing (UDPS) has been used to detect minority variants within HIV-1 populations. Some aspects of the quality and reproducibility of UDPS have been previously evaluated, but comprehensive studies are still needed.

Principal Finding

In this study the UDPS technology (FLX platform) was evaluated by analyzing a 120 base pair fragment of the HIV-1 pol gene from plasma samples from two patients and artificial mixtures of molecular clones. UDPS was performed using an optimized experimental protocol and an in-house data cleaning strategy. Nine samples and mixtures were analyzed and the average number of reads per sample was 19,404 (range 8,858–26,846). The two patient plasma samples were analyzed twice and quantification of viral variants was found to be highly repeatable for variants representing >0.27% of the virus population, whereas some variants representing 0.11–0.27% were detected in only one of the two UDPS runs. Bland-Altman analysis showed that a repeated measurement would have a 95% likelihood to lie approximately within ±0.5 log₁₀ of the initial estimate. A similar level of agreement was observed for variant frequency estimates in forward vs. reverse sequencing direction, but here the agreement was higher for common variants than for rare variants. UDPS following PCR amplification with alternative primers indicated that some variants may be incorrectly quantified due to primer-related selective amplification. Finally, the in vitro recombination rate during PCR was evaluated using artificial mixtures of clones and was found to be low. The most abundant in vitro recombinant represented 0.25% of all UDPS reads.

Conclusion

This study demonstrates that this UDPS protocol results in low experimental noise and high repeatability, which is relevant for future research and clinical use of the UDPS technology. The low rate of in vitro recombination suggests that this UDPS system can be used to study genetic variants and mutational linkage.