Length Bias Research Articles

Editor's Choice

A comparison of tumour and host prognostic factors in screen-detected vs non screen-detected colorectal cancer: a contemporaneous study. Mansouri D, McMillan DC, McIlveen E, Crighton EM, Morrison DS, Horgan PG.1 Current guidelines generally recommend screening for colorectal cancer from the age of 50, as there is compelling evidence that the benefits outweigh the risks. Screening can either prevent cancer by early polyp removal or improve disease-specific survival due to detection of early tumours. As prognosis in colorectal cancer is directly related to stage, the beneficiary effect of screening is by no means a surprise. I have been wondering for a long time, however, whether improved survival is also related to a different biological entity of screen-detected (SD) tumours, with better prognosis than non-screen-detected (NSD) cancer. After all, it might be possible that we are (also) looking at the effect of length-time bias (i.e. artificially improved cancer outcome due to the identification of indolent slow growing tumours). The study in this issue by Mansouri et al. assesses for the first time both host-factors and tumour-phenotype, in addition to tumour stage and site, within the context of a colorectal cancer screening programme. Compared to NSD tumours (both non-responders to screening and patients with interval tumours), patients with SD tumours had less advanced T-stage and less evidence of venous invasion. They had also less evidence of an elevated preoperative systemic inflammatory response. When adjustment was made for stage, however, the two key features representing phenotypically more aggressive tumours (i.e. venous invasion and poor differentiation) did not achieve statistical significance. Subgroup analysis comparing SD tumours with (more aggressive) interval tumours also did not show difference in prognostic factors when adjusted for stage. The conclusion was therefore drawn that the inherent biological characteristics of SD tumours do not differ from those of NSD disease. I agree, although it should be realized that the present study represents a population setting where compliance to the screening programme was just over 50%, and only a minority of interval tumours was detected, which might have influenced the results. So, although slightly reassured, I am still hoping for more studies which address this fascinating subject.

Syllidae mitochondrial gene order is unusually variable for Annelida

Complete mitochondrial genomes of five syllids (Streptosyllis sp., Eusyllis blomstrandi, Myrianida brachycephala, Typosyllis antoni and Typosyllis sp.) have been obtained using Illumina sequencing. Together with two previous studied taxa (Ramisyllis multicaudata and Trypanobia cryptica), the analysed sequences represent most of the main lineages within the family Syllidae (Anoplosyllinae, Eusyllinae, Autolytinae and Syllinae). The genomic features, gene order and phylogenetic relationships are examined. Unusual for annelids, syllid mitochondrial genomes are highly variable in their gene order. Considering genomic features, such as length, skewness, gene content, and codon bias, most similar to the rest of annelids are the genomes of E. blomstrandi and M. brachycephala, while Streptosyllis sp. and the analysed sylline taxa (R. multicaudata, T. cryptica, T. antoni and Typosyllis sp.) are the most dissimilar. Two methionine tRNA's (trnM) have been found in T. antoni and Typosyllis sp. The mt genomes of these latter taxa are the longest with numerous non-coding regions. The 13 protein coding genes, as well as the rRNA's are used to perform phylogenetic analyses that recovered the relationships within the family explored before by previous authors. The gene order in Syllidae shows very different patterns. E. blomstrandi and M. prolifera show a similar pattern to the one found in Pleistoannelida; however this might have changed at least twice within Syllidae: in Streptosyllis sp. and within Syllinae. All analysed Syllinae show different gene orders, thereby illustrating more variability as all other pleistoannelids analysed so far. The information provided herein allows a more accurate reconstruction of the possible evolutionary scenarios in Syllidae.

Benchmarking of the Oxford Nanopore MinION sequencing for quantitative and qualitative assessment of cDNA populations.

To assess the performance of the Oxford Nanopore Technologies MinION sequencing platform, cDNAs from the External RNA Controls Consortium (ERCC) RNA Spike-In mix were sequenced. This mix mimics mammalian mRNA species and consists of 92 polyadenylated transcripts with known concentration. cDNA libraries were generated using a template switching protocol to facilitate the direct comparison between different sequencing platforms. The MinION performance was assessed for its ability to sequence the cDNAs directly with good accuracy in terms of abundance and full length. The abundance of the ERCC cDNA molecules sequenced by MinION agreed with their expected concentration. No length or GC content bias was observed. The majority of cDNAs were sequenced as full length. Additionally, a complex cDNA population derived from a human HEK-293 cell line was sequenced on an Illumina HiSeq 2500, PacBio RS II and ONT MinION platforms. We observed that there was a good agreement in the measured cDNA abundance between PacBio RS II and ONT MinION (rpearson = 0.82, isoforms with length more than 700bp) and between Illumina HiSeq 2500 and ONT MinION (rpearson = 0.75). This indicates that the ONT MinION can sequence quantitatively both long and short full length cDNA molecules.

Estimating patient specific uncertainty parameters for adaptive treatment re-planning in proton therapy using in vivo range measurements and Bayesian inference: application to setup and stopping power errors

In proton therapy, quantification of the proton range uncertainty is important to achieve dose distribution compliance. The promising accuracy of prompt gamma imaging (PGI) suggests the development of a mathematical framework using the range measurements to convert population based estimates of uncertainties into patient specific estimates with the purpose of plan adaptation. We present here such framework using Bayesian inference. The sources of uncertainty were modeled by three parameters: setup bias m, random setup precision r and water equivalent path length bias u. The evolution of the expectation values E(m), E(r) and E(u) during the treatment was simulated. The expectation values converged towards the true simulation parameters after 5 and 10 fractions, for E(m) and E(u), respectively. E(r) settle on a constant value slightly lower than the true value after 10 fractions. In conclusion, the simulation showed that there is enough information in the frequency distribution of the range errors measured by PGI to estimate the expectation values and the confidence interval of the model parameters by Bayesian inference. The updated model parameters were used to compute patient specific lateral and local distal margins for adaptive re-planning.

Effectiveness of surveillance for hepatocellular carcinoma in clinical practice: A United States cohort

The effectiveness of surveillance for hepatocellular carcinoma (HCC) in reducing cancer related mortality among patients with cirrhosis is largely unknown. The objective of this study was to study the effectiveness of HCC surveillance in the national Veterans Administration (VA) clinical practice. We conducted a retrospective cohort study of patients with HCC during 2005-2010 by reviewing patients' medical records to determine receipt of HCC surveillance in the 2years prior to HCC diagnosis. We determined association of HCC surveillance with overall mortality adjusting for age, risk factors, model for end-stage liver disease (MELD) score, comorbidity index, alpha-fetoprotein levels, healthcare utilization, Barcelona Clinic Liver Cancer (BCLC) stage, and treatment. We accounted for lead and length time biases. Of 887 patients with HCC, only 412 (46.5%) received any surveillance prior to HCC diagnosis. Patients who received surveillance were significantly more likely to have early stage disease HCC (BCLC stage 0/A 27.2% vs. 11.6%) and receive potentially curative (20.9% vs. 11.6%) or palliative (59.2% vs. 45.5%) treatments compared to those without HCC surveillance. Receipt of HCC surveillance was associated with 38% reduction in mortality risk (unadjusted hazard ratios (HR) 0.62, 95% confidence intervals (CI) 0.54-0.71) that declined to 20% (HR 0.80, 95% CI 0.69-0.94) after adjusting for HCC stage and treatment, compared to those without HCC surveillance. Among patients with HCC, pre-diagnosis HCC surveillance is associated with a significant 38% reduction in overall mortality. The reduction in mortality risk with surveillance is mediated via stage migration and receipt of HCC specific treatment. Surveillance for liver cancer leads to earlier detection of cancer and increases chances of getting curative treatment. This ultimately leads to increased longevity in patients with liver cancer.

Surrogate Markers: Lessons from the Next Gen?

The article by Banerjee and colleagues published in this issue of the journal involving a randomized control prevention trial of ursodeoxycholic acid (UDCA) in Barrett esophagus reported a null outcome despite being well designed and executed. Possible reasons for this null outcome are discussed focusing on use of surrogate endpoints in the trial. The trial is especially topical because it comes at a time when there are calls for a Pre-Cancer Genome Atlas (PCGA) for "understanding the earliest molecular and cellular events associated with cancer initiation…" This commentary discusses current concepts in prevention research including branched evolution that leads to therapeutic resistance. Length bias sampling postulates underdiagnosis is due to rapidly progressing disease that is difficult to detect by screening because it progresses to cancer too rapidly and that overdiagnosis is the result of very slowly or nonprogressing disease that is easy to detect by screening because it persists for a lifetime and the patient dies of unrelated causes. Finally, it also explores study designs, including surrogate endpoints in Barrett esophagus trials, and opportunities and pitfalls for a PCGA in the context of high levels of over and underdiagnosis of Barrett esophagus as well as many other cancers and their precursors. Cancer Prev Res; 9(7); 512-7. ©2016 AACRSee related article by Banerjee, et al., p. 528.

A novel approach to detect test-seeking behaviour in the blood donor population: making the invisible visible.

Individuals may donate blood in order to determine their infection status after exposure to an increased infection risk. Such test-seeking behaviour decreases transfusion safety. Instances of test seeking are difficult to substantiate as donors are unlikely to admit to such behaviour. However, manifestation in a population of repeat donors may be determined using statistical inference. Test-seeking donors would be highly motivated to donate following infection risk, influencing the timing of their donation. Donation intervals within 2005-2014 of all Dutch blood donors who acquired syphilis (N = 50), HIV (N = 13), HTLV (N = 4) or HCV (N = 2) were compared to donation intervals of uninfected blood donors (N = 7 327 836) using the Anderson-Darling test. We adjusted for length bias as well as for age, gender and donation type of the infected. Additionally, the power of the proposed method was investigated by simulation. Among the Dutch donors who acquired infection, we found only a non-significant overrepresentation of short donation intervals (P = 0·54). However, we show by simulation that both relatively short and long donation intervals among infected donors can reveal test seeking. The power of the method is >90% if among 69 infected donors >35 (51%) are test seeking, or if among 320 infected donors >90 (30%) are test seeking. We show how statistical analysis may be used to reveal the extent of test seeking in repeat blood donor populations. In the Dutch setting, indications for test-seeking behaviour were not statistically significant. This may, however, be due to the low number of infected individuals.

A robust (re-)annotation approach to generate unbiased mapping references for RNA-seq-based analyses of differential expression across closely related species.

BackgroundRNA-seq based on short reads generated by next generation sequencing technologies has become the main approach to study differential gene expression. Until now, the main applications of this technique have been to study the variation of gene expression in a whole organism, tissue or cell type under different conditions or at different developmental stages. However, RNA-seq also has a great potential to be used in evolutionary studies to investigate gene expression divergence in closely related species.ResultsWe show that the published genomes and annotations of the three closely related Drosophila species D. melanogaster, D. simulans and D. mauritiana have limitations for inter-specific gene expression studies. This is due to missing gene models in at least one of the genome annotations, unclear orthology assignments and significant gene length differences in the different species. A comprehensive evaluation of four statistical frameworks (DESeq2, DESeq2 with length correction, RPKM-limma and RPKM-voom-limma) shows that none of these methods sufficiently accounts for inter-specific gene length differences, which inevitably results in false positive candidate genes. We propose that published reference genomes should be re-annotated before using them as references for RNA-seq experiments to include as many genes as possible and to account for a potential length bias. We present a straight-forward reciprocal re-annotation pipeline that allows to reliably compare the expression for nearly all genes annotated in D. melanogaster.ConclusionsWe conclude that our reciprocal re-annotation of previously published genomes facilitates the analysis of significantly more genes in an inter-specific differential gene expression study. We propose that the established pipeline can easily be applied to re-annotate other genomes of closely related animals and plants to improve comparative expression analyses.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-016-2646-x) contains supplementary material, which is available to authorized users.

A Symmetric Length-Aware Enrichment Test.

Young et al., (2010) showed that due to gene length bias the popular Fisher Exact Test should not be used to study the association between a group of differentially expressed (DE) genes and a specific Gene Ontology (GO) category. Instead they suggest a test where one conditions on the genes in the GO category and draws the pseudo DE expressed genes according to a length-dependent distribution. The same model was presented in a different context by Kazemian et al., (2011) who went on to offer a dynamic programming (DP) algorithm to exactly compute the significance of the proposed test. Here we point out that while valid, the test proposed by these authors is no longer symmetric as Fisher's Exact Test is: one gets different answers if one conditions on the observed GO category than on the DE set. As an alternative we offer a symmetric generalization of Fisher's Exact Test and provide efficient algorithms to evaluate its significance.

Length time bias in surveillance for hepatocellular carcinoma and how to avoid it.

Length time bias is a selection bias which can lead to an overestimation of survival of screening-detected cases caused by the relative excess of slower-growing tumors detected with respect to symptomatic cases. This leads to the incorrect perception that screening improves outcomes when it only selects tumors with a favorable biology. Data regarding this bias in surveillance for hepatocellular carcinoma (HCC) have never been provided. A semi-Markov model was developed to investigate this issue. An exponential tumor growth was applied. During its growth, tumor diagnosis "at surveillance appointments" was made when tumor attained a size equal to or above the size of tumors diagnosed in surveilled patients obtained from pertinent published reports, or "in-between appointments" (due to the development of symptoms) if tumor size attained the size of symptomatic diagnosis, derived from published reports; otherwise the tumor continued to grow until the time horizon had been reached. Tumor doubling time (DT) values were recorded according to the method of diagnosis. In a theoretical cohort of 1000 patients submitted to semiannual surveillance, 72.5% will be diagnosed at a surveillance appointment and 18% because of symptom development, although under surveillance. Patients diagnosed with HCC at a surveillance appointment had a median tumor DT of 100days (interquartile range, 68-143days), whereas those diagnosed because of symptoms had a median DT of 42days (interquartile range, 29-58days) although under surveillance. The surveillance propensity to detect slower-growth tumors is relevant, and practical suggestions to minimize this bias in longitudinal studies are provided.

Controversies in Lung Cancer Screening

There remains an extensive debate over lung cancer screening, with lobbying for and against screening for very compelling reasons. The National Lung Screening Trial, International Early Lung Cancer Program, and other major screening studies favor screening with low-dose CT scans and have shown a reduction in lung cancer–specific mortality. The increasing incidence of lung cancer and the dismal survival rate for advanced disease despite improved multimodality therapy have sparked an interest in the implementation of national lung cancer screening. Concerns over imaging workflow, radiation dose, management of small nodules, overdiagnosis bias, lead-time and length-time bias, emerging new technologies, and cost-effectiveness continue to be debated. The authors address each of these issues as they relate to radiologic practice.

Evaluating the Quantitative Capabilities of Metagenomic Analysis Software

DNA sequencing technologies are applied widely and frequently today to describe metagenomes, i.e., microbial communities in environmental or clinical samples, without the need for culturing them. These technologies usually return short (100-300 base-pairs long) DNA reads, and these reads are processed by metagenomic analysis software that assign phylogenetic composition-information to the dataset. Here we evaluate three metagenomic analysis software (AmphoraNet--a webserver implementation of AMPHORA2--, MG-RAST, and MEGAN5) for their capabilities of assigning quantitative phylogenetic information for the data, describing the frequency of appearance of the microorganisms of the same taxa in the sample. The difficulties of the task arise from the fact that longer genomes produce more reads from the same organism than shorter genomes, and some software assign higher frequencies to species with longer genomes than to those with shorter ones. This phenomenon is called the "genome length bias." Dozens of complex artificial metagenome benchmarks can be found in the literature. Because of the complexity of those benchmarks, it is usually difficult to judge the resistance of a metagenomic software to this "genome length bias." Therefore, we have made a simple benchmark for the evaluation of the "taxon-counting" in a metagenomic sample: we have taken the same number of copies of three full bacterial genomes of different lengths, break them up randomly to short reads of average length of 150 bp, and mixed the reads, creating our simple benchmark. Because of its simplicity, the benchmark is not supposed to serve as a mock metagenome, but if a software fails on that simple task, it will surely fail on most real metagenomes. We applied three software for the benchmark. The ideal quantitative solution would assign the same proportion to the three bacterial taxa. We have found that AMPHORA2/AmphoraNet gave the most accurate results and the other two software were under-performers: they counted quite reliably each short read to their respective taxon, producing the typical genome length bias. The benchmark dataset is available at http://pitgroup.org/static/3RandomGenome-100kavg150bps.fna.

Progress Toward Consensus on Breast Cancer Screening Guidelines and Reducing Screening Harms.

Progress Toward Consensus on Breast Cancer Screening Guidelines and Reducing Screening Harms.

Estimating interevent time distributions from finite observation periods in communication networks.

A diverse variety of processes-including recurrent disease episodes, neuron firing, and communication patterns among humans-can be described using interevent time (IET) distributions. Many such processes are ongoing, although event sequences are only available during a finite observation window. Because the observation time window is more likely to begin or end during long IETs than during short ones, the analysis of such data is susceptible to a bias induced by the finite observation period. In this paper, we illustrate how this length bias is born and how it can be corrected without assuming any particular shape for the IET distribution. To do this, we model event sequences using stationary renewal processes, and we formulate simple heuristics for determining the severity of the bias. To illustrate our results, we focus on the example of empirical communication networks, which are temporal networks that are constructed from communication events. The IET distributions of such systems guide efforts to build models of human behavior, and the variance of IETs is very important for estimating the spreading rate of information in networks of temporal interactions. We analyze several well-known data sets from the literature, and we find that the resulting bias can lead to systematic underestimates of the variance in the IET distributions and that correcting for the bias can lead to qualitatively different results for the tails of the IET distributions.

Supercoupling effect in short-channel ultrathin fully depleted silicon-on-insulator transistors

Supercoupling effect prevents the simultaneous formation of inversion and accumulation channels at the two interfaces of ultrathin silicon-on-insulator metal-oxide-semiconductor field-effect transistor. Our work highlights that short-channel effects enhance supercoupling and turn it into a two-dimensional mechanism. The lateral influence of source and drain terminals is evidenced with experimental data and examined by 2D numerical simulations which reveal the roles of gate length and drain bias. The critical Si-film thickness, below which supercoupling arises, is significantly increased in short-channel transistors. The impact of back-gate and drain bias, BOX thickness, and quantum effects is documented and practical applications are discussed.

Screening for nasopharyngeal cancer.

Nasopharyngeal cancer is endemic in a few well-defined populations. The prognosis for advanced nasopharyngeal cancer is poor, but early-stage disease is curable and a high survival rate can be achieved. Screening for early-stage disease could lead to improved outcomes. Epstein-Barr virus (EBV) serology and nasopharyngoscopy are most commonly used for screening. The efficacy and true benefit of screening remain uncertain due to potential selection, lead-time and length-time biases. To determine the effectiveness of screening of asymptomatic individuals by EBV serology and/or nasopharyngoscopy in reducing the mortality of nasopharyngeal cancer compared to no screening. To assess the impact of screening for nasopharyngeal cancer on incidence, survival, adverse effects, cost-effectiveness and quality of life. The Cochrane Ear, Nose and Throat Disorders Group (CENTDG) Trials Search Co-ordinator searched the CENTDG Trials Register; Central Register of Controlled Trials (CENTRAL 2015, Issue 6); PubMed; EMBASE; CINAHL; Web of Science; Clinicaltrials.gov; ICTRP and additional sources for published and unpublished trials. The date of the search was 6 July 2015. Randomised controlled trials (RCT) and controlled clinical trials (CCT) evaluating screening for nasopharyngeal cancer versus no screening. Randomisation either by clusters or individuals was acceptable. We used the standard methodological procedures expected by The Cochrane Collaboration. Our primary outcome measure was nasopharyngeal cancer-specific mortality. Secondary outcomes were incidence of nasopharyngeal cancer by stage and histopathological classification at diagnosis, survival (two-year, three-year, five-year and 10-year), harms of screening (physical and psychosocial), quality of life (via validated tools such as the SF-36 and patient satisfaction), cost-effectiveness and all-cause mortality. We identified no trials that met the review inclusion criteria. We retrieved 31 full-text studies for further investigation following the search. However, none met the eligibility criteria for a RCT or CCT investigation on the efficacy of screening for nasopharyngeal cancer. No data from RCTs or CCTs are available to allow us to determine the efficacy of screening for nasopharyngeal cancer, or the cost-effectiveness and cost-benefit of a screening strategy. High-quality studies with long-term follow-up of mortality and cost-effectiveness are needed.

Alternative splicing acting as a bridge in evolution.

Alternative splicing (AS) regulates diverse cellular and developmental functions through alternative protein structures of different isoforms. Alternative exons dominate AS in vertebrates; however, very little is known about the extent and function of AS in lower eukaryotes. To understand the role of introns in gene evolution, we examined AS from a green algal and five fungal genomes using a novel EST-based gene-modeling algorithm (COMBEST). AS from each genome was classified with COMBEST that maps EST sequences to genomes to build gene models. Various aspects of AS were analyzed through statistical methods. The interplay of intron 3n length, phase, coding property, and intron retention (RI) were examined with Chi-square testing. With 3 to 834 times EST coverage, we identified up to 73% of AS in intron-containing genes and found preponderance of RI among 11 types of AS. The number of exons, expression level, and maximum intron length correlated with number of AS per gene (NAG), and intron-rich genes suppressed AS. Genes with AS were more ancient, and AS was conserved among fungal genomes. Among stopless introns, non-retained introns (NRI) avoided, but major RI preferred 3n length. In contrast, stop-containing introns showed uniform distribution among 3n, 3n+1, and 3n+2 lengths. We found a clue to the intron phase enigma: it was the coding function of introns involved in AS that dictates the intron phase bias. Majority of AS is non-functional, and the extent of AS is suppressed for intron-rich genes. RI through 3n length, stop codon, and phase bias bridges the transition from functionless to functional alternative isoforms.

Four Principles to Consider Before Advising Women on Screening Mammography

This article reviews four important screening principles applicable to screening mammography in order to facilitate informed choice. The first principle is that screening may help, hurt, or have no effect. In order to reduce mortality and mastectomy rates, screening must reduce the rate of advanced disease, which likely has not happened. Through overdiagnosis, screening produces substantial harm by increasing both lumpectomy and mastectomy rates, which offsets the often-promised benefit of less invasive therapy. Next, all-cause mortality is the most reliable way to measure the efficacy of a screening intervention. Disease-specific mortality is biased due to difficulties in attribution of cause of death and to increased mortality due to overdiagnosis and the resulting overtreatment with radiotherapy and chemotherapy. To enhance participation, the benefit from screening is often presented in relative instead of absolute terms. Third, some screening statistics must be interpreted with caution. Increased survival time and the percentage of early-stage tumors at detection sound plausible, but are affected by lead-time and length biases. In addition, analyses that only include women who attend screening cannot reliably correct for selection bias. The final principle is that accounting for tumor biology is important for accurate estimates of lead time, and the potential benefit from screening. Since “early detection” is actually late in a tumor's lifetime, the time window when screen detection might extend a woman's life is narrow, as many tumors that can form metastases will already have done so. Instead of encouraging screening mammography, physicians should help women make an informed decision as with any medical intervention.

Comparison of HLA-A and HLA-B ligandomes

Aim Advances in mass spectrometry and HLA protein availability have allowed for very a detailed look at the self-peptides naturally presented by HLA. To date, there is considerable variation among studies such that a comparison of the ligands presented by HLA-A and HLA-B has been difficult. Here we set out to characterize the self-ligandome of 5 HLA-A and 3 HLA-B molecules representing various peptide binding supertypes with highly controlled methods - varying only the HLA molecule. Methods HeLa cells were separately transfected with 5 different HLA-A (A1, A2, A3, A11, A24) and 3 HLA-B (B7, B27, B51) as soluble class I gene constructs. HLA-complexes were affinity purified and the eluted ligands were characterized by two-dimensional LCMS. Sequences were identified from the fragment spectra using PEAKS v7.0 at a 1% FDR. Results We find that nonamers are the preferred length for all alleles, however there is wide diversity in the ligand length distributions between allomorphs. For example, HLA-B51 shows a high tolerance for 8-mers while HLA-A11 equally prefers 11mers and nonamers. However there is no overall length bias when comparing HLA-A to HLA-B. For both HLA-A and HLA-B, one or two ligands are typically sampled per protein. However, proteins like GAPDH, alpha-enolase 1, carbamoyl-phosphate synthase 1, and dynenin provide on average approximately 20 ligands per allomorph. For all allomorphs there is an uneven sampling of proteins in the genome with biases shown towards particular proteins. Especially noteworthy is that proteins involved in translation, particularly ribosomal RNA binding proteins, are extensively represented by HLA-A and HLA-B. For both HLA-A and HLA-B there is also a significant enrichment of proteins localized to the lumen of the nucleus including histones, proteins involved in RNA splicing, and factors for biogenesis of ribosomes in the nucleolus. Conclusion The ligands sampled by HLA-A and HLA-B vary in length between the allomorphs in a manner that is highly diverse and independent of locus. Overall, there is little to no difference in the source proteins sampled by HLA-A and HLA-B with all allomorphs showing a ligand sampling bias in RNA binding proteins and nuclear proteins. W. Hildebrand: Scientific/Medical Advisor; Company/Organization; Pure Protein LLC. 5. Employee; Company/Organization; Pure Protein LLC.

Short Tree, Long Tree, Right Tree, Wrong Tree: New Acquisition Bias Corrections for Inferring SNP Phylogenies.

Single nucleotide polymorphisms (SNPs) are useful markers for phylogenetic studies owing in part to their ubiquity throughout the genome and ease of collection. Restriction site associated DNA sequencing (RADseq) methods are becoming increasingly popular for SNP data collection, but an assessment of the best practises for using these data in phylogenetics is lacking. We use computer simulations, and new double digest RADseq (ddRADseq) data for the lizard family Phrynosomatidae, to investigate the accuracy of RAD loci for phylogenetic inference. We compare the two primary ways RAD loci are used during phylogenetic analysis, including the analysis of full sequences (i.e., SNPs together with invariant sites), or the analysis of SNPs on their own after excluding invariant sites. We find that using full sequences rather than just SNPs is preferable from the perspectives of branch length and topological accuracy, but not of computational time. We introduce two new acquisition bias corrections for dealing with alignments composed exclusively of SNPs, a conditional likelihood method and a reconstituted DNA approach. The conditional likelihood method conditions on the presence of variable characters only (the number of invariant sites that are unsampled but known to exist is not considered), while the reconstituted DNA approach requires the user to specify the exact number of unsampled invariant sites prior to the analysis. Under simulation, branch length biases increase with the amount of missing data for both acquisition bias correction methods, but branch length accuracy is much improved in the reconstituted DNA approach compared to the conditional likelihood approach. Phylogenetic analyses of the empirical data using concatenation or a coalescent-based species tree approach provide strong support for many of the accepted relationships among phrynosomatid lizards, suggesting that RAD loci contain useful phylogenetic signal across a range of divergence times despite the presence of missing data. Phylogenetic analysis of RAD loci requires careful attention to model assumptions, especially if downstream analyses depend on branch lengths.