Package 'GenomicSEM'

Output from Genomic SEM 'ldsc' function

Summary statistics file created using the 'sumstats' function

Whether the user wants to provide a different standard error (SE) of the SNP variance than the package default. The default is to use .0005 to reflect the fact that the SNP SE is assumed to be population fixed.

addSNPs automatically uses mclapply to create the S and V matrices in parallel. Sometimes running in parallel can cause memory issues within the computing cores. If this is the case, the parallel argument can be set to FALSE, and addSNPs will create the S and V matrices serially.

addSNPs automatically uses mclapply to create the S and V matrices in parallel. If the user does not provide an argument to the cores option, then addSNPs will automatically use one less than the total number of cores available.

Level of Genomic Control (GC) you want the function to use. The default is 'standard' which adjusts the univariate GWAS standard errors by multiplying them by the square root of the univariate LDSC intercept. Additional options include 'conserv' which corrects standard errors using the univariate LDSC intercept, and 'none' which does not correct the standard errors.

variance covariance matrix of the parameter estimates in S that includes an individual SNP effect

genetic covaraiance matrix including individual SNP effect

A list containing relevant genetic information (e.g., rsID, basepair, A1/A2) to be appended to the output from other functions (e.g., userGWAS)

Output from the multivariable LDSC function of Genomic SEM

Options are either Diagonally Weighted Least Squares ("DWLS"; the default) or Maximum Likelihood ("ML")

The model fit results (e.g., model chi-square, AIC, CFI) from the specified model.

Parameter estimates and sandwich corrected standard errors from the specified model.

Output from Genomic SEM 'ldsc' function

Summary statistics file created using the 'sumstats' function

The estimation method to be used when running the factor model. The options are Diagonally Weighted Least Squares ("DWLS", this is the default) or Maximum Likelihood ("ML")

The number of cores to use on your computer for parallel processing. If no number is provided, the default is to use one less core then is available on your computer

The tolerance to use for matrix inversion.

Whether the user wants to provide a different standard error (SE) of the SNP variance than the package default. The default is to use .0005 to reflect the fact that the SNP SE is assumed to be population fixed.

Whether the function should run using parallel or serial processing. Default = TRUE

Level of Genomic Control (GC) you want the function to use. The default is 'standard' which adjusts the univariate GWAS standard errors by multiplying them by the square root of the univariate LDSC intercept. Additional options include 'conserv' which corrects standard errors using the univariate LDSC intercept, and 'none' which does not correct the standard errors.

Whether the function should use multi-node processing (i.e., MPI). Please note that this should only be used on a computing cluster on which the R package Rmpi is already installed.

Whether the function is being used to estimate a multivariate TWAS using read_fusion output for the SNPs argument.

Whether the function should save the consequent largest Zstatistic difference between the pre and post-smooth matrices.

Output from the multivariable S-LDSC function of Genomic SEM (s_ldsc)

Model to be specified using lavaan notation

Parameters of interest to be examined for enrichment (e.g., factor variances).

What components of the model should be fixed for follow-up enrichment models. Default = "regressions", which will fix all regression parameters.

Optional argument to denote whether all latent variables are standardized using unit variance identification (default = FALSE)

Optional argument to denote whether flanking window annotations should automatically be removed from output (default = TRUE)

Optional argument to denote whether the user wants to use the tau genetic covariance matrices, as opposed to the default zero-order matrices, for estimation of enrichment (default = FALSE)

Optional argument to denote whether the user wants to include the full model output from the genome-wide (i.e., baseline) matrix (default = TRUE)

Optional argument to manually set tolerance for matrix inverison.

Optional argument to manually fix paramters when estimating the model within annotations

A vector of strings which point to munged files for trait you want to include in a Genomic SEM model. the HDL function works with standard munged files

A vector of sample prevalences for dichotomous traits and NA for continous traits

A vector of population prevalences for dichotomous traits and NA for continous traits

String which contains the path to the folder in which the LD matrices used in the analysis are located. Expects LD matirices formated as required by the original HDL software.

Sample size of the reference file, default is 335265

A character vector specifying how the traits should be named in the genetic covariance (S) matrix. These variable names can subsequently be used in later steps for model specification. If no value is provided, the function will automatically name the variables using the generic from of V1-VX.

sting, either "piecewise" which estimates the heritability or genetic covariance locally in chunks across the genome and then sums these estimates, or "jackknife" which uses a genoem wide estiamte and uses a jackknife estimator for the variance of the parameter. defaults to "piecewise" the original HDL implementation is equal to "jackknife"

estimated genetic covariance matrix

variance covariance matrix of the parameter estimates in S

matrix containing the "cross trait intercepts", or the error covariance between traits induced by overlap, in terms of subjects, between the GWASes on which the analyses are based

A vector of strings which point to LDSC munged files for trait you want to include in a Genomic SEM model.

A vector of sample prevalences for dichotomous traits and NA for continous traits

A vector of population prevalences for dichotomous traits and NA for continous traits

String which contains the path to the folder in which the LD scores used in the analysis are located. Expects LD scores formated as required by the original LD score regression software.

String which contains the path to the folder in which the LD score weights used in the analysis are located. Expects LD scores formated as required by the original LD score regression software.

A character vector specifying how the traits should be named in the genetic covariance (S) matrix. These variable names can subsequently be used in later steps for model specification. If no value is provided, the function will automatically name the variables using the generic from of V1-VX.

Logical which indicates wheter the weights are different form the LD scores used for the regression, defaults to FALSE.

number of chromosomes over which the LDSC weights are split, defalts to 22 (Human) but can be switched for other species

Number of blocks to use for the jacknive procedure which is used to estiamte entries in V, higher values will be optimal if you have a large number of variables and also slower, defaults to 200

How you want to name your .log file for ldsc. The default is NULL as the package will automatically name the log based in the file names unless a log file name is provided to this arugment.

Whether you want the package to also output a genetic correlation and sampling correlation matrix. Default is FALSE.

Whether you want the package to estimate LDSC using only even or odd chromosomes by setting select to "ODD" and "EVEN" respectively. It can also be set to a set of numbers, such as c(1,3,10), to run ldsc on a specific chromosome or chromosomes. Default is FALSE, in which case LDSC is estimated using all chromosomes.

Maximum value of the squared Z statistics for a SNP that is considered in the LD-score regression. Default behaviour is to set to the maximum of 80 and N*0.001

estimated genetic covariance matrix

variance covariance matrix of the parameter estimates in S

matrix containing the "cross trait intercepts", or the error covariance between traits induced by overlap, in terms of subjects, between the GWASes on which the analyses are based

a vector contsaining the sample size (for the genetic variances) and the geometric mean of sample sizes (i.e. sqrt(N1,N2)) between two samples for the covariances

number of SNPs used to compute the LD scores with.

A vector of parameter values from an unconstrained structural equation model, where focal parameters are estimated freely in each group.

A vector of parameter values from a constrained structural equation model, where focal parameters are constrained to be equal across groups.

A list containing the variances, in each group, of the variables in the usermodel() results column lhs

A list containing the variances, in each group, of the variables in the usermodel() results column rhs

Output from Genomic SEM multivariable LDSC

Summary statistics file created using the sumstats function

Matrix of LD information across the SNPs. If only independent SNPs are being provided a matrix of 0s can be entered. Note that the function requires that A1 and A2 be included in the LD matrix column names (e.g., rs12345_A_T)

User provided SE of the SNP variance for entry in the V matrix. If no number is provided the package defaults to using .0005 to reflect a practically fixed population value taken from a reference panel

List of rsIDs if the user wishes to subset out a set of SNPs from a full set of summary statistics

A vector of file names, files must be located in the working directory, or a path must be provided.

A file of SNPs with A1, A2 and rsID used to allign alleles across traits. We suggest using an (UNZIPPED) file of HAPMAP3 SNPs with some basic cleaning applied (e.g., MHC region removed) that is supplied and created by the original LD score regression developers and available here: https://data.broadinstitute.org/alkesgroup/LDSCORE/w_hm3.snplist.bz2:

A vector of trait names which will be used as names for the munged files

A vector of sample size

Numeric value which is used as a lower bound for inputation quality (INFO)

Numeric value used as a lower bound for minor allel frequency

Optional list detailing which columns represent, SNP, MAF, etc. e.g. list(SNP=my_snp_column)

Indicates whether munge should process the summary statistics files in parallel or serial fashion. Default is TRUE, indicating that it will run in parallel.

Indicates how many cores to use when running in parallel. The default is NULL, in which case munge will use 1 less than the total number of cores available in the local environment.

Indicates whether existing .sumstats.gz files should be overwritten

List of FUSION files.

What names to use when naming the beta and SE columns for each trait

Vector specifying whether each trait is binary (TRUE) or continuous (FALSE)

The sample size to use when backing out the betas and SEs from FUSION Z-statistics. This should reflect the sum of effective sample size for binary traits

Whether you want to use the permutation test p-values from FUSION. Default = FALSE

A vector of file names which point to LDSC munged files for trait you want to include.

A vector of sample prevalences for dichotomous traits and NA for continous traits. Default = NULL.

A vector of population prevalences for dichotomous traits and NA for continous traits. Default = NULL.

A folder (or folders) of partitioned LD scores used as the independent variable in S-LDSC.

A folder of non-partitioned LD scores used as regression weights.

A folder of allele frequency files.

A character vector specifying how the traits should be named. These variable names can subsequently be used in later steps for model specification.

Number of blocks to use for the jacknive procedure which is used to estiamte entries in V, higher values will be optimal if you have a large number of variables and also slower, defaults to 200.

What to name the .log file if you want to overrride default to name file based on file names used as input.

Whether to exclude continuous annotations from S-LDSC estimation.

The zero-order genetic covariance matrices for each annotation.

The zero-order sampling covariance matrices for each annotation.

The tau matrices for each annotation.

The tau sampling covariance matrices for each annotation.

matrix containing the "cross trait intercepts", or the error covariance between traits induced by overlap, in terms of subjects, between the GWASes on which the analyses are based

a vector contsaining the sample size (for the genetic variances) and the geometric mean of sample sizes (i.e. sqrt(N1,N2)) between two samples for the covariances

number of SNPs used to compute the LD scores with.

The proportional size of each annotation relative to the annotation containing all SNPs.

A data.frame that codes flanking window and continuous annotations as 2 and all other annotations as 1. This is used by the 'enrich' function to exclude the flanking window and continuous annotations from enrichment estimates.

a vector of file names, files must be located in the working directory, or a path must be provided.

A reference file of SNPs to keep in your GWAS, one based on 1000 genomes phase 3 is provided.

a vector of trait names which will be used as names for the munged files

a logical vector indicating whether the standard errors in each set of summary statistics is on the logit scale

a logical vector indicating whether the GWAS was for a continuous trait and used OLS (or a LMM)

a logical vector indicating whether the GWAS is a binary outcome with only Z-statistics or was analyzed using a linear probability model i.e. a dichotomous trait using OLS (or a LMM)

A vector of total sample sizes for continuous traits and the sum of effective sample sizes for binary traits

A vector of column names of betas for continuous traits that are known to have been standardized prior to running the GWAS

A vector of sample size

Numeric value which is used as a lower bound for imputation quality (INFO)

Numeric value used as a lower bound for minor allele frequency

Indicates whether insertion-deletion mutations (indels) should be included in your summary statistics. The default is FALSE.

Indicates whether sumstats should process the summary statistics files in parallel or serial fashion. Default is TRUE, indicating that it will run in parallel.

Indicates how many cores to use when running in parallel. The default is NULL, in which case sumstats will use 1 less than the total number of cores available in the local environment.

Indicates whether strand ambiguous SNPs should be removed from output.

Indicates whether SNPs that have missing information for more than half of contributing cohorts, as indicated by missing information in the direction column, should be removed.

Output from Genomic SEM 'ldsc' function

Summary statistics file created using the 'sumstats' function

The estimation method to be used when running the factor model. The options are Diagonally Weighted Least Squares ("DWLS", this is the default) or Maximum Likelihood ("ML")

The user-specified model to use in model estimation using lavaan syntax. The SNP is referred to as 'SNP' in the model.

Whether you want warnings and errors printed for each run. This can take up significant space across all SNPs, but the default is set to TRUE as these warnings may not be safe to ignore.

Whether you want to only output a piece of the model results (e.g., F1 ~ SNP). The argument takes a vector, as multiple pieces of the model result can be output.

Indicates how many cores to use when running in parallel. The default is NULL, in which case sumstats will use 1 less than the total number of cores available in the local environment.

The tolerance to use for matrix inversion.

Whether the user wants to provide a different standard error (SE) of the SNP variance than the package default. The default is to use .0005 to reflect the fact that the SNP SE is assumed to be population fixed.

Whether the function should run using parallel or serial processing. Default = TRUE

Level of Genomic Control (GC) you want the function to use. The default is 'standard' which adjusts the univariate GWAS standard errors by multiplying them by the square root of the univariate LDSC intercept. Additional options include 'conserv' which corrects standard errors using the univariate LDSC intercept, and 'none' which does not correct the standard errors.

Whether the function should use multi-node processing (i.e., MPI). Please note that this should only be used on a computing cluster on which the R package Rmpi is already installed.

Whether the function should save the consequent largest Z-statistic difference between the pre and post-smooth matrices.

Whether the function is being used to estimate a multivariate TWAS using read_fusion output for the SNPs argument.

Optional argument to denote whether all latent variables are standardized using unit variance identification (default = FALSE)

Optional argument to denote whether the measurement model should be fixed across all SNPs (default = TRUE)

Output from the multivariable LDSC function of Genomic SEM

Options are either Diagonally Weighted Least Squares ("DWLS"; the default) or Maximum Likelihood ("ML")

Model to be specified using lavaan notation

Optional argument to denote whether CFI is being requested (default = TRUE). In some cases the estimation of the independent (i.e., Null) model for calculation of CFI can be time consuming. If the funciton seems to be stuck on this step, we would suggest re-running with this option set to FALSE

Optional argument to denote whether all latent variables are standardized using unit variance identification (default = FALSE)

Optional argument to denote whether the user wants the model implied and residual covariance matrix included in the usermodel output (default = FALSE)

Optional argument to denote whether the user wants Genomic SEM to try troubleshooting a model that does not converge by fixing residual variances to be above 0 (default = TRUE)

Optional argument to set lower tolerance for matrix inversion used to produce sadnwich corrected standard errors. (default = FALSE)

The matrix of EFA loadings. Note that the number of columns in this matrix determines how many factors are specifeid in the model.

The LDSC genetic covariance matrix

The EFA standardized loadings cutoff to determine which traits should load on a factor

Whether to apply constraint on all variables to keep residual variances above .001. Default is TRUE.

Whether to specify a bifactor model in which a general factor predicts all included traits and the remaining factors are specifided to be orthogonal of one another.

Whether all variables should load on at least one factor, even if they dont meet the threshold specified using the cutoff argument.

Whether to specify a common factor model.