Package 'cophescan' reference manual

Title:	Adaptation of the Coloc Method for PheWAS
Description:	A Bayesian method for Phenome-wide association studies (PheWAS) that identifies causal associations between genetic variants and traits, while simultaneously addressing confounding due to linkage disequilibrium. For details see Manipur et al (2023) <doi:10.1101/2023.06.29.546856>.
Authors:	Ichcha Manipur [aut, cre], Chris Wallace [aut]
Maintainer:	Ichcha Manipur <[email protected]>
License:	GPL-3
Version:	1.4.1
Built:	2024-10-10 06:46:52 UTC
Source:	CRAN

adjust_priors

Description

adjust fixed priors when nsnps in region is high

Usage

adjust_priors(
  nsnps,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)
adjust_priors(
  nsnps,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)

Arguments

`nsnps`	number of SNPs
`pa`	prior probability that a non-query variant is causally associated with the query trait (cophescan prior), default 3.82e-5
`pc`	prior probability that the query variant is causally associated with the query trait (cophescan prior), default 1.82e-3 (cophescan prior)
`p1`	prior probability a SNP is associated with trait 1, (coloc prior), pc derived by using $pc = p12/p1+p12$ ; use p1, p2, p12 only when pa and pc are unavailable (See vignettes)
`p2`	prior probability a SNP is associated with trait 2, (coloc prior), pa derived by using $pa = p2$
`p12`	prior probability a SNP is associated with both traits, (coloc prior), pc derived by using $pc = p12/p1+p12$

Value

vector of pn, pa and pc adjusted prior probabilities

Average of priors: pnk, pak and pck

Description

Average of priors: pnk, pak and pck

Usage

average_piks(params, nsnps, covar_vec, nits, thin, covar = FALSE)
average_piks(params, nsnps, covar_vec, nits, thin, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`nits`	Number of iterations run in mcmc
`thin`	thinning
`covar`	logical: was the covariate inflormation used? default: False

Value

average pik matrix of priors: pnk, pak and pck

Average of priors: pnk, pak and pck from list (memory intensive)

Description

Average of priors: pnk, pak and pck from list (memory intensive)

Usage

average_piks_list(params, nsnps, covar_vec, nits, thin, covar = FALSE)
average_piks_list(params, nsnps, covar_vec, nits, thin, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`nits`	Number of iterations run in mcmc
`thin`	thinning
`covar`	logical: was the covariate inflormation used? default: False

Value

average pik matrix of priors: pnk, pak and pck

Average of posterior probabilities: Hn, Ha and Hc

Description

Average of posterior probabilities: Hn, Ha and Hc

Usage

average_posterior_prob(
  params,
  lbf_mat,
  nsnps,
  covar_vec,
  nits,
  thin,
  covar = FALSE
)
average_posterior_prob(
  params,
  lbf_mat,
  nsnps,
  covar_vec,
  nits,
  thin,
  covar = FALSE
)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`nits`	Number of iterations run in mcmc
`thin`	thinning
`covar`	logical: was the covariate inflormation used? default: False

Value

matrix with average of all the posterior probabilities: Hn, Ha and Hc

Average of posterior probabilities: Hn, Ha and Hc from list (memory intensive)

Description

Average of posterior probabilities: Hn, Ha and Hc from list (memory intensive)

Usage

average_posterior_prob_list(
  params,
  lbf_mat,
  nsnps,
  covar_vec,
  nits,
  thin,
  covar = FALSE
)
average_posterior_prob_list(
  params,
  lbf_mat,
  nsnps,
  covar_vec,
  nits,
  thin,
  covar = FALSE
)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`nits`	Number of iterations run in mcmc
`thin`	thinning
`covar`	logical: was the covariate inflormation used? default: False

Value

matrix with average of all the posterior probabilities: Hn, Ha and Hc

combine.bf

Description

Calculate posterior probabilities for all the configurations

Usage

combine.bf(lBF_df, pn, pa, pc)
combine.bf(lBF_df, pn, pa, pc)

Arguments

`lBF_df`	dataframe with log bayes factors of hypothesis Ha and Hn: column names should be lBF.Ha and lBF.Hc
`pn`	prior probability that none of the SNPs/variants in the region are associated with the query trait
`pa`	prior probability that a non-query variant is causally associated with the query trait
`pc`	prior probability that the query variant is causally associated with the query trait

Value

named numeric vector of posterior probabilities and bayes factors

Author(s)

Ichcha Manipur

Heatmap of multi-trait cophescan results

Description

Heatmap of multi-trait cophescan results

Usage

cophe_heatmap(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Hc = 0.5,
  thresh_Ha = 0.5,
  ...
)
cophe_heatmap(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Hc = 0.5,
  thresh_Ha = 0.5,
  ...
)

Arguments

`multi.dat`	multi trait cophescan results returned from `cophe.multitrait` or formatted in the same way with multitrait.simplify
`querysnpid`	query variant
`query_trait_names`	names of phenotypes corresponding to the multi.dat results
`thresh_Hc`	Hc threshold to be displayed
`thresh_Ha`	Ha threshold to be displayed
`...`	additional arguments to be passed to pheatmap

Value

heatmap of posterior probabilities of the phentypes above the set threshold

Simulated multi-trait data

Description

Simulated multi-trait data

Usage

data(cophe_multi_trait_data)
data(cophe_multi_trait_data)

Format

list of coloc structred datasets for 24 traits (cophe_multi_trait_data$summ_stat), LD matrix (cophe_multi_trait_data$LD) and the id of the query snp (cophe_multi_trait_data$querysnpid). #' The trait dataset are simulated summary statistics (1000 SNPs) for 10 Hn, 10 Ha and 10 Hc.

cophe_plots showing the Ha and Hc of all traits and labelled above the specified threshold

Description

cophe_plots showing the Ha and Hc of all traits and labelled above the specified threshold

Usage

cophe_plot(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Hc = 0.5,
  thresh_Ha = 0.5,
  beta_p = NULL,
  traits.dat = NULL,
  group_pheno = NULL
)
cophe_plot(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Hc = 0.5,
  thresh_Ha = 0.5,
  beta_p = NULL,
  traits.dat = NULL,
  group_pheno = NULL
)

Arguments

`multi.dat`	multi trait cophescan results returned from cophe.multitrait or multitrait.simplify
`querysnpid`	query variant (only a single variant for PheWAS plots)
`query_trait_names`	list of phenotype names
`thresh_Hc`	Hc threshold to be displayed
`thresh_Ha`	Ha threshold to be displayed
`beta_p`	data.frame (from the `get.beta` function) with four columns : 1. "beta_plot": indicating beta direction (p or n) 2. "beta_plot": -log10(pval) of the queried variant 3. "querysnp" 4. "querytrait".
`traits.dat`	list of multi-trait coloc structured datasets
`group_pheno`	Vector with additional grouping of phenotypes

Value

cophescan plots of Ha and Hc

Predict cophescan hypothesis for tested associations

Description

Predict cophescan hypothesis for tested associations

Usage

cophe.hyp.predict(
  cophe.res,
  grouping.vars = c("querysnp", "querytrait"),
  Hc.cutoff = 0.6,
  Hn.cutoff = 0.2
)
cophe.hyp.predict(
  cophe.res,
  grouping.vars = c("querysnp", "querytrait"),
  Hc.cutoff = 0.6,
  Hn.cutoff = 0.2
)

Arguments

`cophe.res`	results obtained from `cophe.single`, `cophe.susie` or `cophe.multitrait` or data.frame with the following columns: PP.Hn, PP.Hc, PP.Ha, querysnp, querytrait
`grouping.vars`	This is important for results from `cophe.susie` where there are multiple signals. These will be collapsed into one call. If you want to return all signals set this to a single variable eg: grouping.vars = c('querysnp')
`Hc.cutoff`	threshold for PP.Hc above which the associations are called Hc
`Hn.cutoff`	threshold for PP.Hn above which the associations are called Hn

Value

returns dataframe with posterior probabilties of Hn, Hc and Ha with the predicted hypothesis based on the provided cut.offs.

Run cophescan on multiple traits at once

Description

Run cophescan on multiple traits at once

Usage

cophe.multitrait(
  trait.dat,
  querysnpid,
  querytrait.names,
  LDmat = NULL,
  method = "single",
  simplify = FALSE,
  predict.hyp = TRUE,
  Hn.cutoff = 0.2,
  Hc.cutoff = 0.6,
  est.fdr.based.cutoff = FALSE,
  fdr = 0.05,
  ...
)
cophe.multitrait(
  trait.dat,
  querysnpid,
  querytrait.names,
  LDmat = NULL,
  method = "single",
  simplify = FALSE,
  predict.hyp = TRUE,
  Hn.cutoff = 0.2,
  Hc.cutoff = 0.6,
  est.fdr.based.cutoff = FALSE,
  fdr = 0.05,
  ...
)

Arguments

`trait.dat`	Named(traits) list of coloc structured data for k traits (Total number of traits)
`querysnpid`	vector of query variant ids = length(trait.dat), if the same variant
`querytrait.names`	vector of names for the query traits, if the names of the multi.dat list contain the trait names please pass querytrait.names=names(multi.dat)
`LDmat`	LD matrix
`method`	either 'single' for `cophe.single` or 'susie' for `cophe.susie`
`simplify`	if TRUE removes intermediate results from output using 'multitrait.simplify'
`predict.hyp`	if TRUE predicts the hypothesis based on the provided thresholds for pp.Hc and pp.Hn (overrides simplify) using `cophe.hyp.predict`
`Hn.cutoff`	threshold for PP.Hc above which the associations are called Hc
`Hc.cutoff`	threshold for PP.Hc above which the associations are called Hn
`est.fdr.based.cutoff`	if True calculates the Hc.cutoff using 1-mean(PP.Hc)\|PP.Hc > cutoff
`fdr`	fdr threshold to estimate Hc.cutoff
`...`	additional arguments of priors for `cophe.susie` or `cophe.single`

Value

if simplify is False returns multi-trait list of lists, each with:

a summary data.frame of the cophescan results
priors used
querysnp
querytrait

if simplify is TRUE only returns dataframe with posterior probabilties of Hn, Hc and Ha with no intermediate results if predict.hyp is TRUE returns a dataframe with output of simplify and the predicted hypotheses for all associations

Author(s)

Ichcha Manipur

Bayesian cophescan analysis using Approximate Bayes Factors

Description

Bayesian cophescan analysis under single causal variant assumption

Usage

cophe.single(
  dataset,
  querysnpid,
  querytrait,
  MAF = NULL,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)
cophe.single(
  dataset,
  querysnpid,
  querytrait,
  MAF = NULL,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)

Arguments

`dataset`	a list with specifically named elements defining the query trait dataset to be analysed.
`querysnpid`	Id of the query variant, (id in dataset$snp)
`querytrait`	Query trait name
`MAF`	Minor allele frequency vector
`pa`	prior probability that a non-query variant is causally associated with the query trait (cophescan prior), default 3.82e-5
`pc`	prior probability that the query variant is causally associated with the query trait (cophescan prior), default 1.82e-3 (cophescan prior)
`p1`	prior probability a SNP is associated with trait 1, (coloc prior), pc derived by using $pc = p12/p1+p12$ ; use p1, p2, p12 only when pa and pc are unavailable (See vignettes)
`p2`	prior probability a SNP is associated with trait 2, (coloc prior), pa derived by using $pa = p2$
`p12`	prior probability a SNP is associated with both traits, (coloc prior), pc derived by using $pc = p12/p1+p12$

Details

This function calculates posterior probabilities of different causal variant configurations under the assumption of a single causal variant for each trait.

If regression coefficients and variances are available, it calculates Bayes factors for association at each SNP. If only p values are available, it uses an approximation that depends on the SNP's MAF and ignores any uncertainty in imputation. Regression coefficients should be used if available. Find more input data structure details in the coloc package

Value

a list of two data.frames:

summary is a vector giving the number of SNPs analysed, and the posterior probabilities of Hn (no shared causal variant), Ha (two distinct causal variants) and Hc (one common causal variant)
results is an annotated version of the input data containing log Approximate Bayes Factors and intermediate calculations, and the posterior probability SNP.PP.Hc of the SNP being causal for the shared signal if Hc is true. This is only relevant if the posterior support for Hc in summary is convincing.

Author(s)

Ichcha Manipur

Examples

library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
res.single <- cophe.single(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
summary(res.single)
library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
res.single <- cophe.single(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
summary(res.single)

cophe.single.lbf

Description

Calculate log bayes factors for each hypothesis (Single causal variant assumption)

Usage

cophe.single.lbf(dataset, querysnpid, querytrait, MAF = NULL)
cophe.single.lbf(dataset, querysnpid, querytrait, MAF = NULL)

Arguments

`dataset`	a list with specifically named elements defining the query trait dataset to be analysed.
`querysnpid`	Id of the query variant, (id in dataset$snp)
`querytrait`	Query trait name
`MAF`	Minor allele frequency vector

Value

data frame with log bayes factors for Hn and Ha hypotheses

Author(s)

Ichcha Manipur

Examples

library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
res.single.lbf <- cophe.single.lbf(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
res.single.lbf
library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
res.single.lbf <- cophe.single.lbf(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
res.single.lbf

run `cophe.susie` using susie to detect separate signals

Description

Check if a variant causally associated in one trait might be causal in another trait

Usage

cophe.susie(
  dataset,
  querysnpid,
  querytrait,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL,
  susie.args = list()
)
cophe.susie(
  dataset,
  querysnpid,
  querytrait,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL,
  susie.args = list()
)

Arguments

`dataset`	either a list with specifically named elements defining the dataset to be analysed. (see check_dataset)
`querysnpid`	Id of the query variant
`querytrait`	Query trait name
`pa`	prior probability that a non-query variant is causally associated with the query trait (cophescan prior), default 3.82e-5
`pc`	prior probability that the query variant is causally associated with the query trait (cophescan prior), default 1.82e-3
`p1`	prior probability a SNP is associated with trait 1, (coloc prior), pc derived by using $pc = p12/p1+p12$ ; use p1, p2, p12 only when pa and pc are unavailable (See vignettes)
`p2`	prior probability a SNP is associated with trait 2, (coloc prior), pa derived by using $pa = p2$
`p12`	prior probability a SNP is associated with both traits, (coloc prior), pc derived by using $pc = p12/p1+p12$
`susie.args`	a named list of additional arguments to be passed to runsusie

Value

a list, containing elements

summary a data.table of posterior probabilities of each global hypothesis, one row per pairwise comparison of signals from the two traits
results a data.table of detailed results giving the posterior probability for each snp to be jointly causal for both traits assuming Hc is true. Please ignore this column if the corresponding posterior support for H4 is not high.
priors a vector of the priors used for the analysis

Author(s)

Ichcha Manipur

Examples

library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
query_trait_1$LD <- cophe_multi_trait_data$LD
res.susie <- cophe.susie(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
summary(res.susie)
library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
querysnpid <- cophe_multi_trait_data$querysnpid
query_trait_1$LD <- cophe_multi_trait_data$LD
res.susie <- cophe.susie(query_trait_1, querysnpid = querysnpid, querytrait='Trait_1')
summary(res.susie)

cophe.susie.lbf

Description

Calculate log bayes factors for each hypothesis (SuSIE - multiple causal variant assumption)

Usage

cophe.susie.lbf(
  dataset,
  querysnpid,
  querytrait,
  switch = TRUE,
  susie.args = list(),
  MAF = NULL
)
cophe.susie.lbf(
  dataset,
  querysnpid,
  querytrait,
  switch = TRUE,
  susie.args = list(),
  MAF = NULL
)

Arguments

`dataset`	a list with specifically named elements defining the query trait dataset to be analysed.
`querysnpid`	Id of the query variant, (id in dataset$snp)
`querytrait`	Query trait name
`switch`	Set switch=TRUE to obtain single BF when credible sets not found with SuSIE
`susie.args`	a named list of additional arguments to be passed to runsusie
`MAF`	Minor allele frequency vector

Value

data frame with log bayes factors for Hn and Ha hypotheses

Author(s)

Ichcha Manipur

Examples

library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
query_trait_1$LD <- cophe_multi_trait_data$LD
querysnpid <- cophe_multi_trait_data$querysnpid
res.susie.lbf <- cophe.susie.lbf(query_trait_1, querysnpid = querysnpid,
                                  querytrait='Trait_1', switch=T)
res.susie.lbf
library(cophescan)
data(cophe_multi_trait_data)
query_trait_1 <- cophe_multi_trait_data$summ_stat[['Trait_1']]
query_trait_1$LD <- cophe_multi_trait_data$LD
querysnpid <- cophe_multi_trait_data$querysnpid
res.susie.lbf <- cophe.susie.lbf(query_trait_1, querysnpid = querysnpid,
                                  querytrait='Trait_1', switch=T)
res.susie.lbf

Extract beta and p-values of queried variant

Description

Extract beta and p-values of queried variant

Usage

get_beta(traits.dat, querysnpid, querytrait)
get_beta(traits.dat, querysnpid, querytrait)

Arguments

`traits.dat`	list of coloc structured dataset
`querysnpid`	vector of querysnpid
`querytrait`	vector of querytrait names

Value

data.frame with one column named beta_plot: indicating beta direction (n/p) and another column named pval_plot with -log10(pval) of the queried variant

Calculation of the posterior prob of Hn, Ha and Hc

Description

Calculation of the posterior prob of Hn, Ha and Hc

Usage

get_posterior_prob(params, lbf_mat, nsnps, covar_vec, covar = FALSE)
get_posterior_prob(params, lbf_mat, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: should the covariate inflormation be used? default: False

Value

posterior prob of Hn, Ha and Hc

Estimate the Hc.cutoff for the required FDR

Description

Estimate the Hc.cutoff for the required FDR

Usage

Hc.cutoff.fdr(ppHc, fdr = 0.05, return_plot = TRUE)
Hc.cutoff.fdr(ppHc, fdr = 0.05, return_plot = TRUE)

Arguments

`ppHc`	a vector containing the PP.Hc (the posterior probability of causal association) of all tests
`fdr`	FDR default: 0.05
`return_plot`	default: TRUE, plot the fdr estimated at the different Hc.cutoff

Value

the Hc.cutoff value for the specified FDR, if return_plot is True returns a plot showing the FDR calculated at different Hc thresholds

hypothesis.priors

Description

Estimate priors for each hypothesis

Usage

hypothesis.priors(nsnps, pn, pa, pc)
hypothesis.priors(nsnps, pn, pa, pc)

Arguments

`nsnps`	number of SNPs
`pn`	prior probability that none of the SNPs/variants in the region are associated with the query trait
`pa`	prior probability that a non-query variant is causally associated with the query trait
`pc`	prior probability that the query variant is causally associated with the query trait

Value

hypotheses priors

Author(s)

Ichcha Manipur

dnorm for alpha

Description

dnorm for alpha

Usage

logd_alpha(a, alpha_mean = -10, alpha_sd = 0.5)
logd_alpha(a, alpha_mean = -10, alpha_sd = 0.5)

Arguments

`a`	current alpha
`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha

Value

log dnorm

dgamma for beta

Description

dgamma for beta

Usage

logd_beta(b, beta_shape = 2, beta_scale = 2)
logd_beta(b, beta_shape = 2, beta_scale = 2)

Arguments

`b`	current beta
`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta

Value

log dgamma

dgamma for gamma

Description

dgamma for gamma

Usage

logd_gamma(g, gamma_shape = 2, gamma_scale = 2)
logd_gamma(g, gamma_shape = 2, gamma_scale = 2)

Arguments

`g`	current gamma
`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

log dgamma

Log likelihood calculation

Description

Log likelihood calculation

Usage

loglik(params, lbf_mat, nsnps, covar_vec, covar = FALSE)
loglik(params, lbf_mat, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: should the covariate inflormation be used? default: False

Value

logpost flog of the posteriors

Log posterior calculation

Description

Log posterior calculation

Usage

logpost(params, lbf_mat, nsnps, covar_vec, covar = FALSE)
logpost(params, lbf_mat, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: should the covariate inflormation be used? default: False

Value

logpost flog of the posteriors

Calculate log priors

Description

Calculate log priors

Usage

logpriors(
  params,
  covar = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)
logpriors(
  params,
  covar = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`covar`	logical: Should the covariate inflormation be used? default: False
`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha
`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta
`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

log priors

logsum

Description

Internal function, logsum Function directly taken from coloc This function calculates the log of the sum of the exponentiated logs taking out the max, i.e. insuring that the sum is not Inf

Usage

logsum(x)
logsum(x)

Arguments

`x`	numeric vector

Value

max(x) + log(sum(exp(x - max(x))))

Log sum

Description

Log sum

Usage

logsumexp(x)
logsumexp(x)

Arguments

`x`	vector of log scale values to be added

Value

log sum of input

Run the hierarchical mcmc model to infer priors

Description

Run the hierarchical mcmc model to infer priors

Usage

metrop_run(
  lbf_mat,
  nsnps,
  covar_vec,
  covar = FALSE,
  nits = 10000L,
  thin = 1L,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)
metrop_run(
  lbf_mat,
  nsnps,
  covar_vec,
  covar = FALSE,
  nits = 10000L,
  thin = 1L,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)

Arguments

`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: Should the covariate inflormation be used? default: False
`nits`	Number of iterations run in mcmc
`thin`	thinning
`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha
`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta
`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

named list of log likelihood (ll) and parameters: alpha, beta and gamma

Simplifying the output obtained from `cophe.multitrait`, `cophe.single` or `cophe.susie`

Description

Simplifying the output obtained from cophe.multitrait, cophe.single or cophe.susie

Usage

multitrait.simplify(multi.dat, only_BF = FALSE)
multitrait.simplify(multi.dat, only_BF = FALSE)

Arguments

`multi.dat`	output obtained from `cophe.multitrait`, `cophe.single` or `cophe.susie`
`only_BF`	return only bayes factors and not posterior probabilities (default=FALSE)

Value

dataframe with posterior probabilties of Hn, Hc and Ha

Initiate parameters alpha, beta and gamma

Description

Initiate parameters alpha, beta and gamma

Usage

pars_init(
  covar = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)
pars_init(
  covar = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)

Arguments

`covar`	logical: Should the covariate inflormation be used? default: False
`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha
`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta
`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

params $\alpha$ , $\beta$ and $\gamma$

Conversion of parameters alpha, beta and gamma to pnk, pak and pck

Description

Conversion of parameters alpha, beta and gamma to pnk, pak and pck

Usage

pars2pik(params, nsnps, covar_vec, covar = FALSE)
pars2pik(params, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: should the covariate information be used? default: False

Value

pik matrix of priors: pnk, pak and pck

per.snp.priors

Description

Estimate per snp priors

Usage

per.snp.priors(
  nsnps,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)
per.snp.priors(
  nsnps,
  pa = 3.82e-05,
  pc = 0.00182,
  p1 = NULL,
  p2 = NULL,
  p12 = NULL
)

Arguments

`nsnps`	number of SNPs
`pa`	prior probability that a non-query variant is causally associated with the query trait (cophescan prior), default 3.82e-5
`pc`	prior probability that the query variant is causally associated with the query trait (cophescan prior), default 1.82e-3 (cophescan prior)
`p1`	prior probability a SNP is associated with trait 1, (coloc prior), pc derived by using $pc = p12/p1+p12$ ; use p1, p2, p12 only when pa and pc are unavailable (See vignettes)
`p2`	prior probability a SNP is associated with trait 2, (coloc prior), pa derived by using $pa = p2$
`p12`	prior probability a SNP is associated with both traits, (coloc prior), pc derived by using $pc = p12/p1+p12$

Value

priors at the query variant

Author(s)

Ichcha Manipur

List of priors: pn, pa and pc over all iterations

Description

List of priors: pn, pa and pc over all iterations

Usage

piks(params, nsnps, covar_vec, covar = FALSE)
piks(params, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: was the covariate inflormation used? default: False

Value

List of priors (len: iterations): pnk, pak and pck

Plot region Manhattan for a trait highlighting the queried variant

Description

Plot region Manhattan for a trait highlighting the queried variant

Usage

plot_trait_manhat(trait.dat, querysnpid, alt.snpid = NULL)
plot_trait_manhat(trait.dat, querysnpid, alt.snpid = NULL)

Arguments

`trait.dat`	dataset used as input for running cophescan
`querysnpid`	the id of the causal variant as present in trait.dat$snp, plotted in red
`alt.snpid`	the id of the other variants as a vector to be plotted, plotted in blue

Value

regional manhattan plot

List of posterior probabilities: Hn, Ha and Hc over all iterations

Description

List of posterior probabilities: Hn, Ha and Hc over all iterations

Usage

posterior_prob(params, lbf_mat, nsnps, covar_vec, covar = FALSE)
posterior_prob(params, lbf_mat, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: was the covariate inflormation used? default: False

Value

List of posterior probabilties (len: iterations): Hn, Ha and Hc

Prepare data for plotting

Description

Prepare data for plotting

Usage

prepare_plot_data(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Ha = 0.5,
  thresh_Hc = 0.5,
  hmp = FALSE,
  cophe.plot = TRUE
)
prepare_plot_data(
  multi.dat,
  querysnpid,
  query_trait_names,
  thresh_Ha = 0.5,
  thresh_Hc = 0.5,
  hmp = FALSE,
  cophe.plot = TRUE
)

Arguments

`multi.dat`	multi trait cophescan results returned from cophe.multitrait or multitrait.simplify
`querysnpid`	query variant
`query_trait_names`	vector of names of the query traits
`thresh_Ha`	Ha threshold to be displayed
`thresh_Hc`	Hc threshold to be displayed
`hmp`	return for heatmap
`cophe.plot`	default: TRUE, return for `cophe_plot`

Value

plot list

Proposal distribution

Description

Proposal distribution

Usage

propose(params, propsd = 0.5)
propose(params, propsd = 0.5)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`propsd`	Standard deviation for the proposal

Value

vector : proposal

Run the hierarchical Metropolis Hastings model to infer priors

Description

Run the hierarchical Metropolis Hastings model to infer priors

Usage

run_metrop_priors(
  multi.dat,
  covar = FALSE,
  covar_vec = NULL,
  is_covar_categorical = FALSE,
  nits = 10000,
  thin = 1,
  posterior = FALSE,
  avg_pik = TRUE,
  avg_posterior = TRUE,
  pik = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)
run_metrop_priors(
  multi.dat,
  covar = FALSE,
  covar_vec = NULL,
  is_covar_categorical = FALSE,
  nits = 10000,
  thin = 1,
  posterior = FALSE,
  avg_pik = TRUE,
  avg_posterior = TRUE,
  pik = FALSE,
  alpha_mean = -10,
  alpha_sd = 0.5,
  beta_shape = 2,
  beta_scale = 2,
  gamma_shape = 2,
  gamma_scale = 2
)

Arguments

`multi.dat`	matrix of bf values, rows=traits, named columns=("lBF.Ha","lBF.Hc","nsnps")
`covar`	whether to include covariates
`covar_vec`	vector of covariates
`is_covar_categorical`	only two categories supported (default=FALSE) - Experimental
`nits`	number of iterations
`thin`	burnin
`posterior`	default: FALSE, estimate posterior probabilities of the hypotheses
`avg_pik`	default: FALSE, estimate the average of the pik
`avg_posterior`	default: FALSE, estimate the average of the posterior probabilities of the hypotheses
`pik`	default: FALSE, inferred prior probabilities
`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha
`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta
`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

List containing the posterior distribution of the parameters alpha, beta, gamma (if covariate included) and the loglikelihood

if avg_posterior=TRUE matrix with average of all the posterior probabilities of Hn, Ha and Hc

if avg_pik=TRUE matrix with average of all the priors: pn, pa and pc

data, nits and thin contain the input data, number of iterations and burnin respectively specified for the hierarchical model

sample alpha

Description

sample alpha

Usage

sample_alpha(alpha_mean = -10, alpha_sd = 0.5)
sample_alpha(alpha_mean = -10, alpha_sd = 0.5)

Arguments

`alpha_mean`	prior for the mean of alpha
`alpha_sd`	prior for the standard deviation of alpha

Value

sample from rnorm for $\alpha$

sample beta

Description

sample beta

Usage

sample_beta(beta_shape = 2, beta_scale = 2)
sample_beta(beta_shape = 2, beta_scale = 2)

Arguments

`beta_shape`	prior for the shape (gamma distibution) of beta
`beta_scale`	prior for the scale of beta

Value

sample from rgamma for $\beta$

sample gamma

Description

sample gamma

Usage

sample_gamma(gamma_shape = 2, gamma_scale = 2)
sample_gamma(gamma_shape = 2, gamma_scale = 2)

Arguments

`gamma_shape`	prior for the shape (gamma distibution) of gamma
`gamma_scale`	prior for the scale of gamma

Value

sample from rgamma for $\gamma$

print the summary of results from cophescan single or susie

Description

print the summary of results from cophescan single or susie

Usage

## S3 method for class 'cophe'
summary(object, ...)
## S3 method for class 'cophe'
summary(object, ...)

Arguments

`object`	Result from either `cophe.susie` or `cophe.single`
`...`	additional arguments affecting the summary produced.

Value

log bayes and posterior probabilities

Target distribution

Description

Target distribution

Usage

target(params, lbf_mat, nsnps, covar_vec, covar = FALSE)
target(params, lbf_mat, nsnps, covar_vec, covar = FALSE)

Arguments

`params`	Vector of parameters: $\alpha$ , $\beta$ and $\gamma$
`lbf_mat`	matrix of log bayes factors: lBF.Ha and lBF.Hc
`nsnps`	number of snps
`covar_vec`	Vector of the covariate
`covar`	logical: Should the covariate inflormation be used? default: False

Value

target

Package 'cophescan'

Help Index

The 'cophescan' package.

Description

adjust_priors

Description

Usage

Arguments

Value

Average of priors: pnk, pak and pck

Description

Usage

Arguments

Value

Average of priors: pnk, pak and pck from list (memory intensive)

Description

Usage

Arguments

Value

Average of posterior probabilities: Hn, Ha and Hc

Description

Usage

Arguments

Value

Average of posterior probabilities: Hn, Ha and Hc from list (memory intensive)

Description

Usage

Arguments

Value

combine.bf

Description

Usage

Arguments

Value

Author(s)

Heatmap of multi-trait cophescan results

Description

Usage

Arguments

Value

Simulated multi-trait data

Description

Usage

Format

cophe_plots showing the Ha and Hc of all traits and labelled above the specified threshold

Description

Usage

Arguments

Value

See Also

Predict cophescan hypothesis for tested associations

Description

Usage

Arguments

Value

See Also

Run cophescan on multiple traits at once

Description

Usage

Arguments

Value

Author(s)

Bayesian cophescan analysis using Approximate Bayes Factors

Description

Usage

Arguments

Details

Value

Author(s)

Examples

cophe.single.lbf

Description

Usage

Arguments

Value

Author(s)

See Also

Examples

run cophe.susie using susie to detect separate signals

Description

run `cophe.susie` using susie to detect separate signals