compute.graphical.scores: A function to compute SNP-pair scores for network plots of...

In mnodzenski/epistasisGA: An R package to identify multi-snp effects in nuclear family studies using the GADGETS method

A function to compute SNP-pair scores for network plots of results.

Description

This function returns a data.table of graphical SNP-pair scores for use in network plots of GADGETS results.

Usage

compute.graphical.scores(
  results.list,
  preprocessed.list,
  score.type = "logsum",
  pval.thresh = 0.05,
  n.permutes = 10000,
  n.different.snps.weight = 2,
  n.both.one.weight = 1,
  weight.function.int = 2,
  recessive.ref.prop = 0.75,
  recode.test.stat = 1.64,
  bp.param = bpparam(),
  null.mean.vec.list = NULL,
  null.sd.vec.list = NULL
)

Arguments

results.list	A list of length d, where d is the number of chromosome sizes to be included in the network plot. Each element of the list should be a data.table from combine.islands for a given chromosome size. Each data.table in the list should be subset to only include those chromosomes whose fitness scores are high enough to contribute to the network plot. The selection of the chromosomes that contribute to these plots is at the analyst's discretion. We have found success in just using the top 10 scoring chromosomes, and also by restricting attention to those chromosomes that exceed the 95th percentile of the maxima observed after running GADGETS on data-sets permuted under a global, no-association null. For the latter, see also function global.test.
preprocessed.list	The list output by preprocess.genetic.data run on the observed data.
score.type	A character string specifying the method for aggregating SNP-pair scores across chromosome sizes. Options are 'max', 'sum', or 'logsum', defaulting to 'logsum'. For a given SNP-pair, it's graphical score will be the score.type of all graphical scores of chromosomes containing that pair across chromosome sizes. The choice of 'logsum' rather than 'sum' may be useful in cases where there are multiple risk-sets, and one is found much more frequently. However, it may be of interest to examine plots using both score.type approaches.
pval.thresh	A numeric value between 0 and 1 specifying the epistasis test p-value threshold for a chromosome to contribute to the network. Any chromosomes with epistasis p-value greater than pval.thresh will not contribute to network plots. The argument defaults to 0.05. It must be <= 0.6.
n.permutes	The number of permutations on which to base the epistasis tests. Defaults to 10000.
n.different.snps.weight	The number by which the number of different SNPs between a case and complement/unaffected sibling is multiplied in computing the family weights. Defaults to 2.
n.both.one.weight	The number by which the number of SNPs equal to 1 in both the case and complement/unaffected sibling is multiplied in computing the family weights. Defaults to 1.
weight.function.int	An integer used to assign family weights. Specifically, we use weight.function.int in a function that takes the weighted sum of the number of different SNPs and SNPs both equal to one as an argument, denoted as x, and returns a family weight equal to weight.function.int^x. Defaults to 2.
recessive.ref.prop	The proportion to which the observed proportion of informative cases with the provisional risk genotype(s) will be compared to determine whether to recode the SNP as recessive. Defaults to 0.75.
recode.test.stat	For a given SNP, the minimum test statistic required to recode and recompute the fitness score using recessive coding. Defaults to 1.64.
bp.param	The BPPARAM argument to be passed to bplapply. See BiocParallel::bplapply for more details.
null.mean.vec.list	(experimental) A list, equal in length to results.list, where the i^th element of the list is the vector of null means (stored in the 'null.mean.sd.info.rds') corresponding to the d (chromosome size) used to generate the results stored in the i^th element of results.list. This only needs to be specified if based on the experimental E-GADGETS method, and otherwise can be left at its default.
null.sd.vec.list	(experimental) A list, equal in length to results.list, where the i^th element of the list is the vector of null standard deviations (stored in the 'null.mean.sd.info.rds') corresponding to the d (chromosome size) used to generate the results stored in the i^th element of results.list. This only needs to be specified if based on the experimental E-GADGETS method, and otherwise can be left at its default.

Value

A list of two elements:

pair.scores

A data.table containing SNP-pair graphical scores, where the first four columns represent SNPs and the fifth column (pair.score) is the graphical SNP-pair score.

snp.scores

A data.table containing individual SNP graphical scores, where the first two columns represent SNPs and the third column (snp.score) is the graphical SNP score.

Examples

data(case)
data(dad)
data(mom)
data(snp.annotations)
set.seed(1400)

# preprocess data
target.snps <- c(1:3, 30:32, 60:62, 85)
preprocessed.list <- preprocess.genetic.data(as.matrix(case[, target.snps]),
                        father.genetic.data = as.matrix(dad[ , target.snps]),
                        mother.genetic.data = as.matrix(mom[ , target.snps]),
                     ld.block.vec = c(3, 3, 3, 1))
## run GA for observed data

#observed data chromosome size 2
run.gadgets(preprocessed.list, n.chromosomes = 5, chromosome.size = 2,
       results.dir = 'tmp_2',
       cluster.type = 'interactive',
       registryargs = list(file.dir = 'tmp_reg', seed = 1500),
       generations = 2, n.islands = 2, island.cluster.size = 1,
       n.migrations = 0)
 combined.res2 <- combine.islands('tmp_2',
                     snp.annotations[ target.snps, ], preprocessed.list, 2)
 unlink('tmp_reg', recursive = TRUE)

 #observed data chromosome size 3
 run.gadgets(preprocessed.list, n.chromosomes = 5,
       chromosome.size = 3, results.dir = 'tmp_3',
       cluster.type = 'interactive',
       registryargs = list(file.dir = 'tmp_reg', seed = 1500),
       generations = 2, n.islands = 2, island.cluster.size = 1,
       n.migrations = 0)
 combined.res3 <- combine.islands('tmp_3', snp.annotations[ target.snps, ],
                                  preprocessed.list, 2)
 unlink('tmp_reg', recursive = TRUE)

## create list of results

final.results <- list(combined.res2[1:3, ], combined.res3[1:3, ])

 ## compute edge scores
 edge.dt <- compute.graphical.scores(final.results,
                                     preprocessed.list,
                                     pval.thresh = 0.5)

lapply(c("tmp_2", "tmp_3"), unlink, recursive = TRUE)

mnodzenski/epistasisGA documentation built on Jan. 17, 2023, 7:07 p.m.