Estimating population haplotype frequencies from pooled DNA samples using PHASE algorithm

Computational Haplotype Inference from Pooled Samples

Computationally inferring the identities and their relative frequencies from pooled samples that are whole-genome or segmentally genotyped or sequenced (e.g., using next-generation sequencing) in a pool is useful for population genetics analysis. To carry out such analysis, one needs to understand basics of how to use high-performance computing (HPC) facilities and the specifics of corresponding computational tools. Here, we describe the basic knowledge and step-by-step usage of a number of tools for haplotype inference on genotyping or next-generation sequencing data.

An EM algorithm based on an internal list for estimating haplotype distributions of rare variants from pooled genotype data

Background

Pooling is a cost effective way to collect data for genetic association studies, particularly for rare genetic variants. It is of interest to estimate the haplotype frequencies, which contain more information than single locus statistics. By viewing the pooled genotype data as incomplete data, the expectation-maximization (EM) algorithm is the natural algorithm to use, but it is computationally intensive. A recent proposal to reduce the computational burden is to make use of database information to form a list of frequently occurring haplotypes, and to restrict the haplotypes to come from this list only in implementing the EM algorithm. There is, however, the danger of using an incorrect list, and there may not be enough database information to form a list externally in some applications.

Results

We investigate the possibility of creating an internal list from the data at hand. One way to form such a list is to collapse the observed total minor allele frequencies to “zero” or “at least one”, which is shown to have the desirable effect of amplifying the haplotype frequencies. To improve coverage, we propose ways to add and remove haplotypes from the list, and a benchmarking method to determine the frequency threshold for removing haplotypes. Simulation results show that the EM estimates based on a suitably augmented and trimmed collapsed data list (ATCDL) perform satisfactorily. In two scenarios involving 25 and 32 loci respectively, the EM-ATCDL estimates outperform the EM estimates based on other lists as well as the collapsed data maximum likelihood estimates.

Conclusions

The proposed augmented and trimmed CD list is a useful list for the EM algorithm to base upon in estimating the haplotype distributions of rare variants. It can handle more markers and larger pool size than existing methods, and the resulting EM-ATCDL estimates are more efficient than the EM estimates based on other lists.

Fast and accurate haplotype frequency estimation for large haplotype vectors from pooled DNA data

Background

Typically, the first phase of a genome wide association study (GWAS) includes genotyping across hundreds of individuals and validation of the most significant SNPs. Allelotyping of pooled genomic DNA is a common approach to reduce the overall cost of the study. Knowledge of haplotype structure can provide additional information to single locus analyses. Several methods have been proposed for estimating haplotype frequencies in a population from pooled DNA data.

Results

We introduce a technique for haplotype frequency estimation in a population from pooled DNA samples focusing on datasets containing a small number of individuals per pool (2 or 3 individuals) and a large number of markers. We compare our method with the publicly available state-of-the-art algorithms HIPPO and HAPLOPOOL on datasets of varying number of pools and marker sizes. We demonstrate that our algorithm provides improvements in terms of accuracy and computational time over competing methods for large number of markers while demonstrating comparable performance for smaller marker sizes. Our method is implemented in the "Tree-Based Deterministic Sampling Pool" (TDSPool) package which is available for download at http://www.ee.columbia.edu/~anastas/tdspool.

Conclusions

Using a tree-based determinstic sampling technique we present an algorithm for haplotype frequency estimation from pooled data. Our method demonstrates superior performance in datasets with large number of markers and could be the method of choice for haplotype frequency estimation in such datasets.

Estimating the effect of SNP genotype on quantitative traits from pooled DNA samples

Background

Studies to detect associations between DNA markers and traits of interest in humans and livestock benefit from increasing the number of individuals genotyped. Performing association studies on pooled DNA samples can provide greater power for a given cost. For quantitative traits, the effect of an SNP is measured in the units of the trait and here we propose and demonstrate a method to estimate SNP effects on quantitative traits from pooled DNA data.

Methods

To obtain estimates of SNP effects from pooled DNA samples, we used logistic regression of estimated allele frequencies in pools on phenotype. The method was tested on a simulated dataset, and a beef cattle dataset using a model that included principal components from a genomic correlation matrix derived from the allele frequencies estimated from the pooled samples. The performance of the obtained estimates was evaluated by comparison with estimates obtained using regression of phenotype on genotype from individual samples of DNA.

Results

For the simulated data, the estimates of SNP effects from pooled DNA are similar but asymptotically different to those from individual DNA data. Error in estimating allele frequencies had a large effect on the accuracy of estimated SNP effects. For the beef cattle dataset, the principal components of the genomic correlation matrix from pooled DNA were consistent with known breed groups, and could be used to account for population stratification. Correctly modeling the contemporary group structure was essential to achieve estimates similar to those from individual DNA data, and pooling DNA from individuals within groups was superior to pooling DNA across groups. For a fixed number of assays, pooled DNA samples produced results that were more correlated with results from individual genotyping data than were results from one random individual assayed from each pool.

Conclusions

Use of logistic regression of allele frequency on phenotype makes it possible to estimate SNP effects on quantitative traits from pooled DNA samples. With pooled DNA samples, genotyping costs are reduced, and in cases where trait records are abundant this approach is promising to obtain SNP associations for marker-assisted selection.

Estimating haplotype frequencies by combining data from large DNA pools with database information

We assume that allele frequency data have been extracted from several large DNA pools, each containing genetic material of up to hundreds of sampled individuals. Our goal is to estimate the haplotype frequencies among the sampled individuals by combining the pooled allele frequency data with prior knowledge about the set of possible haplotypes. Such prior information can be obtained, for example, from a database such as HapMap. We present a Bayesian haplotyping method for pooled DNA based on a continuous approximation of the multinomial distribution. The proposed method is applicable when the sizes of the DNA pools and/or the number of considered loci exceed the limits of several earlier methods. In the example analyses, the proposed model clearly outperforms a deterministic greedy algorithm on real data from the HapMap database. With a small number of loci, the performance of the proposed method is similar to that of an EM-algorithm, which uses a multinormal approximation for the pooled allele frequencies, but which does not utilize prior information about the haplotypes. The method has been implemented using Matlab and the code is available upon request from the authors.

Estimating population haplotype frequencies from pooled SNP data using incomplete database information

MOTIVATION

Information about haplotype structures gives a more detailed picture of genetic variation between individuals than single-locus analyses. Databases that contain the most frequent haplotypes of certain populations are developing rapidly (e.g. the HapMap database for single-nucleotide polymorphisms in humans). Utilization of such prior information about the prevailing haplotype structures makes it possible to estimate the haplotype frequencies also from large DNA pools. When genetic material from dozens of individuals is pooled together and analysed in a single genotyping, the overall number of genotypings and the costs of the genetic studies are reduced.

RESULTS

A Bayesian model for estimating the haplotypes and their frequencies from pooled allelic observations is introduced. The model combines an idea of using database information for haplotype estimation with a computationally efficient multinormal approximation. In addition, the model treats the number and structures of the unknown haplotypes as random variables whose joint posterior distribution is estimated. The results on real human data from the HapMap database show that the proposed method provides significant improvements over the existing methods.

AVAILABILITY

A reversible-jump Markov chain Monte Carlo algorithm for analysing the model is implemented in a program called Hippo (Haplotype estimation under incomplete prior information using pooled observations). For comparisons, an approximate expectation-maximization algorithm (EM-algorithm) that utilizes database information about the existing haplotypes is implemented in a program called AEML. The source codes written in C (using GNU Scientific Library) are available at www.iki.fi/ approximately mpirinen.

CONTACT

matti.pirinen@iki.fi