Skip to content
Snippets Groups Projects
Commit 9130f5a4 authored by Hanna  JULIENNE's avatar Hanna JULIENNE
Browse files

update documentation

parent f6f688c3
No related branches found
No related tags found
1 merge request!8Improve performance report and made function to generate LD from plink more flexible
...@@ -18,14 +18,15 @@ What is RAISS ? ...@@ -18,14 +18,15 @@ What is RAISS ?
RAISS is a python package to impute missing SNP summary statistics from RAISS is a python package to impute missing SNP summary statistics from
neighboring SNPs in linkage desiquilibrium. neighboring SNPs in linkage desiquilibrium.
The statistical model used to make the imputation is described in :cite:`Pasaniuc2014` and in :cite:`lee2013dist`. The statistical model used to make the imputation is described in :cite:`Pasaniuc2014`, :cite:`lee2013dist` and :cite:`Julienne2019`.
The implementation and performances of RAISS are described in :cite:`Julienne2019`.
The imputation execution time is optimized by precomputing Linkage desiquilibrium between SNPs. The imputation execution time is optimized by precomputing Linkage desiquilibrium between SNPs.
Dependencies Dependencies
============ ============
RAISS requires plink version 1.9 : `<https://www.cog-genomics.org/plink2>`_ RAISS requires plink version 1.9 for the precomputation of LD: `<https://www.cog-genomics.org/plink2>`_
Installation Installation
============ ============
...@@ -34,6 +35,11 @@ Installation ...@@ -34,6 +35,11 @@ Installation
pip3 install git+https://gitlab.pasteur.fr/statistical-genetics/raiss.git pip3 install git+https://gitlab.pasteur.fr/statistical-genetics/raiss.git
Alternatively, RAISS is available as a docker container:
https://quay.io/repository/biocontainers/raiss?tab=tags
Precomputation of LD-correlation Precomputation of LD-correlation
================================= =================================
...@@ -45,6 +51,9 @@ To limit the number of SNP pairs, the LD is computed between pairs of ...@@ -45,6 +51,9 @@ To limit the number of SNP pairs, the LD is computed between pairs of
SNPs in a approximately LD-independent regions. For an european ancestry, you can use SNPs in a approximately LD-independent regions. For an european ancestry, you can use
the region defined by :cite:`Berisa2015` that are provided in the package data folder. the region defined by :cite:`Berisa2015` that are provided in the package data folder.
For other ancestries, we recommend computing approximately LD-independent regions using
`bigsnpr <https://privefl.github.io/bigsnpr/reference/snp_ldsplit.html>`_ (:cite:`privee2022`).
To compute the LD you need to specify a reference panel splitted by chromosomes To compute the LD you need to specify a reference panel splitted by chromosomes
(bed, fam and bim formats of plink, see `PLINK formats <https://www.cog-genomics.org/plink2/formats>`_ ) (bed, fam and bim formats of plink, see `PLINK formats <https://www.cog-genomics.org/plink2/formats>`_ )
...@@ -59,6 +68,9 @@ To compute the LD you need to specify a reference panel splitted by chromosomes ...@@ -59,6 +68,9 @@ To compute the LD you need to specify a reference panel splitted by chromosomes
ld_folder_out = "/mnt/atlas/PCMA/WKD_Hanna/impute_for_jass/berisa_ld_block" ld_folder_out = "/mnt/atlas/PCMA/WKD_Hanna/impute_for_jass/berisa_ld_block"
raiss.LD.generate_genome_matrices(, ...) raiss.LD.generate_genome_matrices(, ...)
Since 2021, LD can also be specified as scipy sparse matrix (.npz), the index must be provided in a separate file (one id by line)
Input format: Input format:
============= =============
...@@ -89,13 +101,15 @@ Output ...@@ -89,13 +101,15 @@ Output
The raiss package outputs imputed GWAS files in the tabular format: The raiss package outputs imputed GWAS files in the tabular format:
+-------------+----------+------------+------------+---------+-------+----------+ +-------------+----------+------------+------------+---------+-------+----------+---------------+
| snp_ids | position | Ref_allele | Alt_allele | z_score | Var | ld_score | | rsID | pos | A0 | A1 | Z | Var | ld_score | imputation_R2 |
+=============+==========+============+============+=========+=======+==========+ +=============+==========+============+============+=========+=======+==========+===============+
| rs3802985 | 198510 | T | C | 0.334 | -1.0 | -1.0 | | rs3802985 | 198510 | T | C | 0.334 | -1.0 | inf | 2.0 |
+-------------+----------+------------+------------+---------+-------+----------+ +-------------+----------+------------+------------+---------+-------+----------+---------------+
| rs111876722 | 201922 | C | T | 0.297 | 0.16 | 5.412 | | rs111876722 | 201922 | C | T | 0.297 | 0.09 | 5.412 | 0.91578 |
+-------------+----------+------------+------------+---------+-------+----------+ +-------------+----------+------------+------------+---------+-------+----------+---------------+
Variance is set to -1 for variants present in the input dataset
Optimizing RAISS parameter for your data Optimizing RAISS parameter for your data
======================================== ========================================
......
@article{Julienne2019,
author = {Julienne, Hanna and Shi, Huwenbo and Pasaniuc, Bogdan and Aschard, Hugues},
title = "{RAISS: robust and accurate imputation from summary statistics}",
journal = {Bioinformatics},
volume = {35},
number = {22},
pages = {4837-4839},
year = {2019},
month = {06},
abstract = "{Multi-trait analyses using public summary statistics from genome-wide association studies (GWASs) are becoming increasingly popular. A constraint of multi-trait methods is that they require complete summary data for all traits. Although methods for the imputation of summary statistics exist, they lack precision for genetic variants with small effect size. This is benign for univariate analyses where only variants with large effect size are selected a posteriori. However, it can lead to strong p-value inflation in multi-trait testing. Here we present a new approach that improve the existing imputation methods and reach a precision suitable for multi-trait analyses.We fine-tuned parameters to obtain a very high accuracy imputation from summary statistics. We demonstrate this accuracy for variants of all effect sizes on real data of 28 GWAS. We implemented the resulting methodology in a python package specially designed to efficiently impute multiple GWAS in parallel.The python package is available at: https://gitlab.pasteur.fr/statistical-genetics/raiss, its accompanying documentation is accessible here http://statistical-genetics.pages.pasteur.fr/raiss/.Supplementary data are available at Bioinformatics online.}",
issn = {1367-4803},
doi = {10.1093/bioinformatics/btz466},
url = {https://doi.org/10.1093/bioinformatics/btz466},
eprint = {https://academic.oup.com/bioinformatics/article-pdf/35/22/4837/30706731/btz466.pdf},
}
@article{privee2022,
author = {Privé, Florian},
title = "{Optimal linkage disequilibrium splitting}",
journal = {Bioinformatics},
volume = {38},
number = {1},
pages = {255-256},
year = {2021},
month = {07},
abstract = "{A few algorithms have been developed for splitting the genome in nearly independent blocks of linkage disequilibrium. Due to the complexity of this problem, these algorithms rely on heuristics, which makes them suboptimal.Here, we develop an optimal solution for this problem using dynamic programming.This is now implemented as function snp\_ldsplit as part of R package bigsnpr.Supplementary data are available at Bioinformatics online.}",
issn = {1367-4803},
doi = {10.1093/bioinformatics/btab519},
url = {https://doi.org/10.1093/bioinformatics/btab519},
eprint = {https://academic.oup.com/bioinformatics/article-pdf/38/1/255/41891000/btab519.pdf},
}
@article{Pasaniuc2014, @article{Pasaniuc2014,
abstract = {MOTIVATION Imputation using external reference panels (e.g. 1000 Genomes) is a widely used approach for increasing power in genome-wide association studies and meta-analysis. Existing hidden Markov models (HMM)-based imputation approaches require individual-level genotypes. Here, we develop a new method for Gaussian imputation from summary association statistics, a type of data that is becoming widely available. RESULTS In simulations using 1000 Genomes (1000G) data, this method recovers 84{\%} (54{\%}) of the effective sample size for common ({\textgreater}5{\%}) and low-frequency (1-5{\%}) variants [increasing to 87{\%} (60{\%}) when summary linkage disequilibrium information is available from target samples] versus the gold standard of 89{\%} (67{\%}) for HMM-based imputation, which cannot be applied to summary statistics. Our approach accounts for the limited sample size of the reference panel, a crucial step to eliminate false-positive associations, and it is computationally very fast. As an empirical demonstration, we apply our method to seven case-control phenotypes from the Wellcome Trust Case Control Consortium (WTCCC) data and a study of height in the British 1958 birth cohort (1958BC). Gaussian imputation from summary statistics recovers 95{\%} (105{\%}) of the effective sample size (as quantified by the ratio of [Formula: see text] association statistics) compared with HMM-based imputation from individual-level genotypes at the 227 (176) published single nucleotide polymorphisms (SNPs) in the WTCCC (1958BC height) data. In addition, for publicly available summary statistics from large meta-analyses of four lipid traits, we publicly release imputed summary statistics at 1000G SNPs, which could not have been obtained using previously published methods, and demonstrate their accuracy by masking subsets of the data. We show that 1000G imputation using our approach increases the magnitude and statistical evidence of enrichment at genic versus non-genic loci for these traits, as compared with an analysis without 1000G imputation. Thus, imputation of summary statistics will be a valuable tool in future functional enrichment analyses. AVAILABILITY AND IMPLEMENTATION Publicly available software package available at http://bogdan.bioinformatics.ucla.edu/software/. CONTACT bpasaniuc@mednet.ucla.edu or aprice@hsph.harvard.edu SUPPLEMENTARY INFORMATION Supplementary materials are available at Bioinformatics online.}, abstract = {MOTIVATION Imputation using external reference panels (e.g. 1000 Genomes) is a widely used approach for increasing power in genome-wide association studies and meta-analysis. Existing hidden Markov models (HMM)-based imputation approaches require individual-level genotypes. Here, we develop a new method for Gaussian imputation from summary association statistics, a type of data that is becoming widely available. RESULTS In simulations using 1000 Genomes (1000G) data, this method recovers 84{\%} (54{\%}) of the effective sample size for common ({\textgreater}5{\%}) and low-frequency (1-5{\%}) variants [increasing to 87{\%} (60{\%}) when summary linkage disequilibrium information is available from target samples] versus the gold standard of 89{\%} (67{\%}) for HMM-based imputation, which cannot be applied to summary statistics. Our approach accounts for the limited sample size of the reference panel, a crucial step to eliminate false-positive associations, and it is computationally very fast. As an empirical demonstration, we apply our method to seven case-control phenotypes from the Wellcome Trust Case Control Consortium (WTCCC) data and a study of height in the British 1958 birth cohort (1958BC). Gaussian imputation from summary statistics recovers 95{\%} (105{\%}) of the effective sample size (as quantified by the ratio of [Formula: see text] association statistics) compared with HMM-based imputation from individual-level genotypes at the 227 (176) published single nucleotide polymorphisms (SNPs) in the WTCCC (1958BC height) data. In addition, for publicly available summary statistics from large meta-analyses of four lipid traits, we publicly release imputed summary statistics at 1000G SNPs, which could not have been obtained using previously published methods, and demonstrate their accuracy by masking subsets of the data. We show that 1000G imputation using our approach increases the magnitude and statistical evidence of enrichment at genic versus non-genic loci for these traits, as compared with an analysis without 1000G imputation. Thus, imputation of summary statistics will be a valuable tool in future functional enrichment analyses. AVAILABILITY AND IMPLEMENTATION Publicly available software package available at http://bogdan.bioinformatics.ucla.edu/software/. CONTACT bpasaniuc@mednet.ucla.edu or aprice@hsph.harvard.edu SUPPLEMENTARY INFORMATION Supplementary materials are available at Bioinformatics online.},
archivePrefix = {arXiv}, archivePrefix = {arXiv},
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment