Search

Identity-by-descent (IBD) segments are a useful tool for applications ranging from demographic inference to relationship classification, but most detection methods rely on phasing information and therefore require substantial computation time. As genetic datasets grow, methods for inferring IBD segments that scale well will be critical. We developed IBIS, an IBD detector that locates long regions of allele sharing between unphased individuals, and benchmarked it with Refined IBD, GERMLINE, and TRUFFLE on 3,000 simulated individuals. Phasing these with Beagle 5 takes 4.3 CPU days, followed by either Refined IBD or GERMLINE segment detection in 2.9 or 1.1 h, respectively. By comparison, IBIS finishes in 6.8 min or 7.8 min with IBD2 functionality enabled: speedups of 805-946× including phasing time. TRUFFLE takes 2.6 h, corresponding to IBIS speedups of 20.2-23.3×. IBIS is also accurate, inferring ≥7 cM IBD segments at quality comparable to Refined IBD and GERMLINE. With these segments, IBIS classifies first through third degree relatives in real Mexican American samples at rates meeting or exceeding other methods tested and identifies fourth through sixth degree pairs at rates within 0.0%-2.0% of the top method. While allele frequency-based approaches that do not detect segments can infer relationship degrees faster than IBIS, the fastest are biased in admixed samples, with KING inferring 30.8% fewer fifth degree Mexican American relatives correctly compared with IBIS. Finally, we ran IBIS on chromosome 2 of the UK Biobank dataset and estimate its runtime on the autosomes to be 3.3 days parallelized across 128 cores.

Summary The proportion of samples with one or more close relatives in a genetic dataset increases rapidly with sample size, necessitating relatedness modeling and enabling pedigree-based analyses. Despite this, relatives are generally unreported and current inference methods typically detect only the degree of relatedness of sample pairs and not pedigree relationships. We developed CREST, an accurate and fast method that identifies the pedigree relationships of close relatives. CREST utilizes identity by descent (IBD) segments shared between a pair of samples and their mutual relatives, leveraging the fact that sharing rates among these individuals differ across pedigree configurations. Furthermore, CREST exploits the profound differences in sex-specific genetic maps to classify pairs as maternally or paternally related—e.g., paternal half-siblings—using the locations of autosomal IBD segments shared between the pair. In simulated data, CREST correctly classifies 91.5%–100% of grandparent-grandchild (GP) pairs, 80.0%–97.5% of avuncular (AV) pairs, and 75.5%–98.5% of half-siblings (HS) pairs compared to PADRE’s rates of 38.5%–76.0% of GP, 60.5%–92.0% of AV, 73.0%–95.0% of HS pairs. Turning to the real 20,032 sample Generation Scotland (GS) dataset, CREST identified seven pedigrees with incorrect relationship types or maternal/paternal parent sexes, five of which we confirmed as mistakes, and two with uncertain relationships. After correcting these, CREST correctly determines relationship types for 93.5% of GP, 97.7% of AV, and 92.2% of HS pairs that have sufficient mutual relative data; the parent sex in 100% of HS and 99.6% of GP pairs; and it completes this analysis in 2.8 h including IBD detection in eight threads.

Simulations of close relatives and identical by descent (IBD) segments are common in genetic studies, yet most past efforts have utilized sex averaged genetic maps and ignored crossover interference, thus omitting features known to affect the breakpoints of IBD segments. We developed Ped-sim, a method for simulating relatives that can utilize either sex-specific or sex averaged genetic maps and also either a model of crossover interference or the traditional Poisson model for inter-crossover distances. To characterize the impact of previously ignored mechanisms, we simulated data for all four combinations of these factors. We found that modeling crossover interference decreases the standard deviation of pairwise IBD proportions by 10.4% on average in full siblings through second cousins. By contrast, sex-specific maps increase this standard deviation by 4.2% on average, and also impact the number of segments relatives share. Most notably, using sex-specific maps, the number of segments half-siblings share is bimodal; and when combined with interference modeling, the probability that sixth cousins have non-zero IBD sharing ranges from 9.0 to 13.1%, depending on the sexes of the individuals through which they are related. We present new analytical results for the distributions of IBD segments under these models and show they match results from simulations. Finally, we compared IBD sharing rates between simulated and real relatives and find that the combination of sex-specific maps and interference modeling most accurately captures IBD rates in real data. Ped-sim is open source and available from https://github.com/williamslab/ped-sim., Author summary Simulations are ubiquitous throughout statistical genetics in order to generate data with known properties, enabling tests of inference methods and analyses of real world processes in settings where experimental data are challenging to collect. Simulating genetic data for relatives in a pedigree requires the synthesis of chromosomes parents transmit to their children. These chromosomes form as a mosaic of a given parent’s two chromosomes, with the location of switches between the two parental chromosomes known as crossovers. Detailed information about crossover generation based on real data from humans now exists, including the fact that men and women have overall different rates (women produce ∼1.6 times more crossovers) and that real crossovers are subject to interference—whereby crossovers are further apart from one another than expected under a model that selects their locations randomly. Our new method, Ped-sim, can simulate pedigree data using these less commonly modeled crossover features, and we used it to evaluate the impact of sex-specific rates and interference compared to real data. These comparisons show that both factors shape the amount of DNA two relatives share, and that their inclusion in models of crossover better fit data from real relatives.

As genetic datasets increase in size, the fraction of samples with one or more close relatives grows rapidly, resulting in sets of mutually related individuals. We present DRUID—deep relatedness utilizing identity by descent—a method that works by inferring the identical-by-descent (IBD) sharing profile of an ungenotyped ancestor of a set of close relatives. Using this IBD profile, DRUID infers relatedness between unobserved ancestors and more distant relatives, thereby combining information from multiple samples to remove one or more generations between the deep relationships to be identified. DRUID constructs sets of close relatives by detecting full siblings and also uses an approach to identify the aunts/uncles of two or more siblings, recovering 92.2% of real aunts/uncles with zero false positives. In real and simulated data, DRUID correctly infers up to 10.5% more relatives than PADRE when using data from two sets of distantly related siblings, and 10.7%–31.3% more relatives given two sets of siblings and their aunts/uncles. DRUID frequently infers relationships either correctly or within one degree of the truth, with PADRE classifying 43.3%–58.3% of tenth degree relatives in this way compared to 79.6%–96.7% using DRUID.

Relatedness inference is an essential component of many genetic analyses and popular in consumer genetic testing. Ramstetter et al. evaluate twelve....., Inferring relatedness from genomic data is an essential component of genetic association studies, population genetics, forensics, and genealogy. While numerous methods exist for inferring relatedness, thorough evaluation of these approaches in real data has been lacking. Here, we report an assessment of 12 state-of-the-art pairwise relatedness inference methods using a data set with 2485 individuals contained in several large pedigrees that span up to six generations. We find that all methods have high accuracy (92–99%) when detecting first- and second-degree relationships, but their accuracy dwindles to 76% of relative pairs. Overall, the most accurate methods are Estimation of Recent Shared Ancestry (ERSA) and approaches that compute total IBD sharing using the output from GERMLINE and Refined IBD to infer relatedness. Combining information from the most accurate methods provides little accuracy improvement, indicating that novel approaches, such as new methods that leverage relatedness signals from multiple samples, are needed to achieve a sizeable jump in performance.

The proportion of samples with one or more close relatives in a genetic dataset increases rapidly with sample size, necessitating relatedness modeling and enabling pedigree-based analyses. Despite this, relatives are generally unreported and current inference methods typically detect only the degree of relatedness of sample pairs and not pedigree relationships. We developed CREST, an accurate and fast method that identifies the pedigree relationships of close relatives. CREST utilizes identical by descent (IBD) segments shared between a pair of samples and their mutual relatives, leveraging the fact that sharing rates among these individuals differ across pedigree configurations. Furthermore, CREST exploits the profound differences in sex-specific genetic maps to classify pairs as maternally or paternally related—e.g., paternal half-siblings—using the locations of autosomal IBD segments shared between the pair. In simulated data, CREST correctly classifies 91.5-99.5% of grandparent-grandchild (GP) pairs, 70.5-97.0% of avuncular (AV) pairs, and 79.0-98.0% of half-siblings (HS) pairs compared to PADRE’s rates of 38.5-76.0% of GP, 60.5-92.0% of AV, 73.0-95.0% of HS pairs. Turning to the real 20,032 sample Generation Scotland (GS) dataset, CREST correctly determines the relationship of 99.0% of GP, 85.7% of AV, and 95.0% of HS pairs that have sufficient mutual relative data, completing this analysis in 10.1 CPU hours including IBD detection. CREST’s maternal and paternal relationship inference is also accurate, as it flagged five pairs as incorrectly labeled in the GS pedigrees— three of which we confirmed as mistakes, and two with an uncertain relationship—yielding 99.7% of HS and 93.5% of GP pairs correctly classified.

The sequencing of Neanderthal and Denisovan genomes has yielded many new insights about interbreeding events between extinct hominins and the ancestors of modern humans. While much attention has been paid to the relatively recent gene flow from Neanderthals and Denisovans into modern humans, other instances of introgression leave more subtle genomic evidence and have received less attention. Here, we present a major extension of the ARGweaver algorithm, called ARGweaver-D, which can infer local genetic relationships under a user-defined demographic model that includes population splits and migration events. This Bayesian algorithm probabilistically samples ancestral recombination graphs (ARGs) that specify not only tree topologies and branch lengths along the genome, but also indicate migrant lineages. The sampled ARGs can therefore be parsed to produce probabilities of introgression along the genome. We show that this method is well powered to detect the archaic migration into modern humans, even with only a few samples. We then show that the method can also detect introgressed regions stemming from older migration events, or from unsampled populations. We apply it to human, Neanderthal, and Denisovan genomes, looking for signatures of older proposed migration events, including ancient humans into Neanderthal, and unknown archaic hominins into Denisovans. We identify 3% of the Neanderthal genome that is putatively introgressed from ancient humans, and estimate that the gene flow occurred between 200-300kya. We find no convincing evidence that negative selection acted against these regions. Finally, we predict that 1% of the Denisovan genome was introgressed from an unsequenced, but highly diverged, archaic hominin ancestor. About 15% of these “super-archaic” regions—comprising at least about 4Mb—were, in turn, introgressed into modern humans and continue to exist in the genomes of people alive today., Author summary We present ARGweaver-D, an extension of the ARGweaver algorithm which can be applied under a user-defined demographic model including population splits and migration events. Given genome sequence data from a collection of individuals across multiple closely related populations or subspecies, ARGweaver-D can infer trees describing the genetic relationships among these individuals at every location along the genome, conditional on the demographic model. Like ARGweaver, ARGweaver-D is a Bayesian method, sampling trees from the posterior distribution in order to account for uncertainty. Using simulations, we show that ARGweaver-D can successfully identify regions introgressed from Neanderthals and Denisovans into modern humans. It is also well-powered to detect introgressed regions stemming from older gene-flow events. We apply ARGweaver-D to the genomes of two Neanderthals, a Denisovan, and two African humans. We identify 3% of the Neanderthal genome which is likely derived from gene flow from ancient humans. We also identify about 1% of the Denisovan genome that may be traced to an unsequenced archaic hominin; 15% of these regions were subsequently passed to modern humans. We find no convincing evidence that selection acted against any of these introgressed regions.

As genetic datasets increase in size, the fraction of samples with one or more close relatives grows rapidly, resulting in sets of mutually related individuals. We present DRUID—Deep Relatedness Utilizing Identity by Descent—a method that works by inferring the identical by descent (IBD) sharing profile of an ungenotyped ancestor of a set of close relatives. Using this IBD profile, DRUID infers relatedness between unobserved ancestors and more distant relatives, thereby combining information from multiple samples to remove one or more generations between the deep relationships to be identified. DRUID constructs sets of close relatives by detecting full siblings and also uses a novel approach to identify the aunts/uncles of two or more siblings, recovering 92.2% of real aunts/uncles with zero false positives. In real and simulated data, DRUID correctly infers up to 10.5% more relatives than PADRE when using data from two sets of distantly related siblings, and 10.7–31.3% more relatives given two sets of siblings and their aunts/uncles. DRUID frequently infers relationships either correctly or within one degree of the truth, with PADRE classifying 43.3–58.3% of tenth degree relatives in this way compared to 79.6–96.7% using DRUID.

Objectives Academic and not-for-profit research funders are increasingly requiring that the research they fund must be published open access, with some insisting on publishing with a Creative Commons Attribution (CC BY) licence to allow the broadest possible use. We set out to clarify the open access variants provided by leading medical journals for research in general and industry-funded research in particular, and record the availability of the CC BY licence for commercially funded research. Methods We identified medical journals with a 2015 impact factor of at least 15.0 on 24 May 2017, then excluded from the analysis journals that only publish review articles. Between 29 June 2017 and 26 July 2017, we collected information about each journal’s open access policies from their websites and/or by email contact. We contacted the journals by email again between 6 December 2017 and 2 January 2018 to confirm our findings. Results Thirty-five medical journals publishing original research from 13 publishers were included in the analysis. All 35 journals offered some form of open access with varying embargo periods of up to 12 months. Of these journals, 21 (60%) provided immediate open access with a CC BY licence under certain circumstances (e.g. to specific research funders). Of these 21, 20 only offered a CC BY licence to authors funded by non-commercial organizations and one offered this option to funders who required it. Conclusions Most leading medical journals do not offer to authors reporting commercially funded research an open access licence that allows unrestricted sharing and adaptation of the published material. The journals’ policies are therefore not aligned with open access declarations and guidelines. Commercial research funders lag behind academic funders in the development of mandatory open access policies, and it is time for them to work with publishers to advance the dissemination of the research they fund. Strengths and limitations of this study This manuscript includes a systematic analysis of open access policies of journals with a high impact factor, including society-owned journals, from multiple publishers. The open access policies of all journals analysed were clarified, and confirmation of our findings was received by email from 97% of the contacted journals. Open access policies of the journals and publishers analysed are subject to change, so the information presented here may not be current. By selecting journals with a high impact factor, our analysis does not include prestigious journals from specialized therapy areas and regional or non-English language journals, which may have lower impact factors. Although our study covers only a small number of journals, extending such a manual analysis to a greater number of journals without loss of detail and verification of all results would be cumbersome and inefficient by relying on traditional analysis tools.

Haplotypes are an important resource for a large number of applications in human genetics, but computationally inferred haplotypes are subject to switch errors that decrease their utility. The accuracy of computationally inferred haplotypes increases with sample size, and although ever larger genotypic data sets are being generated, the fact that existing methods require substantial computational resources limits their applicability to data sets containing tens or hundreds of thousands of samples. Here, we present HAPI-UR (haplotype inference for unrelated samples), an algorithm that is designed to handle unrelated and/or trio and duo family data, that has accuracy comparable to or greater than existing methods, and that is computationally efficient and can be applied to 100,000 samples or more. We use HAPI-UR to phase a data set with 58,207 samples and show that it achieves practical runtime and that switch errors decrease with sample size even with the use of samples from multiple ethnicities. Using a data set with 16,353 samples, we compare HAPI-UR to Beagle, MaCH, IMPUTE2, and SHAPEIT and show that HAPI-UR runs 18× faster than all methods and has a lower switch-error rate than do other methods except for Beagle; with the use of consensus phasing, running HAPI-UR three times gives a slightly lower switch-error rate than Beagle does and is more than six times faster. We demonstrate results similar to those from Beagle on another data set with a higher marker density. Lastly, we show that HAPI-UR has better runtime scaling properties than does Beagle so that for larger data sets, HAPI-UR will be practical and will have an even larger runtime advantage. HAPI-UR is available online (see Web Resources).

Although the past decade has seen tremendous progress in our understanding of fine-scale recombination, little is known about non-crossover (NCO) gene conversion. We report the first genome-wide study of NCO events in humans. Using SNP array data from 98 meioses, we identified 103 sites affected by NCO, of which 50/52 were confirmed in sequence data. Overlap with double strand break (DSB) hotspots indicates that most of the events are likely of meiotic origin. We estimate that a site is involved in a NCO at a rate of 5.9 × 10−6/bp/generation, consistent with sperm-typing studies, and infer that tract lengths span at least an order of magnitude. Observed NCO events show strong allelic bias at heterozygous AT/GC SNPs, with 68% (58–78%) transmitting GC alleles (p = 5 × 10−4). Strikingly, in 4 of 15 regions with resequencing data, multiple disjoint NCO tracts cluster in close proximity (∼20–30 kb), a phenomenon not previously seen in mammals. DOI: http://dx.doi.org/10.7554/eLife.04637.001, eLife digest The genetic information inside our cells is stored in the form of chromosomes, which are carefully packaged strands of DNA. Most human cells contain a pair of each chromosome: one inherited from the mother and another from the father. Typically, when a human cell divides, it duplicates all of its chromosomes and then places one copy of each into the two new cells. However, a different process—known as ‘meiosis’—occurs when a human cell divides to make the cells involved in sexual reproduction (i.e., egg cells in females and sperm cells in males). First, the cell duplicates all of its chromosomes as before, but then it pairs the chromosomes originally from the mother with the equivalent chromosomes from the father. These paired chromosomes then swap sections of DNA. Next, the cell divides, and the resulting cells divide again; this produces four new cells that each contain a single, unique copy of every chromosome. In the process of swapping sections of DNA between chromosomes, the DNA molecule inside the chromosome is broken and different sections of DNA are then joined together. This can occur by one of two methods: ‘crossover events’ that produce a final chromosome made up of long sequences from each of the contributing chromosomes; and ‘non-crossover events’, where only a small section of DNA is swapped between the chromosomes. Research has tended to focus on DNA breaks and crossover events. Now, Williams et al. have looked at the genetic sequences transmitted by both parents to 49 humans—revealing information about a total of 98 meioses—and scoured them for evidence of non-crossover events. In addition to finding 103 sites where these events occurred, Williams et al. discovered that non-crossover events are more frequent around sites where crossover events also have a higher frequency. This suggests that the mechanism that initiates non-crossover events is shared with crossovers, and that non-crossover events primarily occur during meiosis. Unexpectedly, in some areas non-crossover events were found close to each other in ‘clusters’, which had not previously been seen in humans. Non-crossover events will only produce an observable change if the chromosomes involved have differences in the sequence of the DNA section that is swapped between them. The number of such variable genetic positions that non-crossover events affect in a generation is roughly the same number as the number of newly generated random mutations to the DNA sequence in a generation. Examining the DNA sequences transferred during non-crossover events also shows that two different types of DNA bases (cytosine and guanine) are more likely to be transmitted by a non-crossover event than are the other two bases (adenine and thymine). This bias indicates that non-crossover events are an important factor in driving genome evolution. In the future, sequencing the entire genome—the total genetic material—of many different people could provide further insights into non-crossover events in humans. DOI: http://dx.doi.org/10.7554/eLife.04637.002

Hapi is a new dynamic programming algorithm that ignores uninformative states and state transitions in order to efficiently compute minimum-recombinant and maximum likelihood haplotypes. When applied to a dataset containing 103 families, Hapi performs 3.8 and 320 times faster than state-of-the-art algorithms. Because Hapi infers both minimum-recombinant and maximum likelihood haplotypes and applies to related individuals, the haplotypes it infers are highly accurate over extended genomic distances., National Institutes of Health (U.S.) (NIH grant 5-T90-DK070069), National Institutes of Health (U.S.) (Grant 5-P01-NS055923), National Science Foundation (U.S.) (Graduate Research Fellowship)

We undertook a two-stage genome-wide association study (GWAS) of Alzheimer's disease (AD) involving over 16,000 individuals, the most powerful AD GWAS to date. In stage 1 (3,941 cases and 7,848 controls), we replicated the established association with the apolipoprotein E (APOE) locus (most significant SNP, rs2075650, P = 1.8 × 10−157) and observed genome-wide significant association with SNPs at two loci not previously associated with the disease: at the CLU (also known as APOJ) gene (rs11136000, P = 1.4 × 10−9) and 5′ to the PICALM gene (rs3851179, P = 1.9 × 10−8). These associations were replicated in stage 2 (2,023 cases and 2,340 controls), producing compelling evidence for association with Alzheimer's disease in the combined dataset (rs11136000, P = 8.5 × 10−10, odds ratio = 0.86; rs3851179, P = 1.3 × 10−9, odds ratio = 0.86).