Package LDM. R topics documented: March 19, Type Package

Transcription

1 Type Package Package LDM March 19, 2018 Title Testing Hypotheses about the Microbiome using an Ordination-based Linear Decomposition Model Version 1.0 Date Depends GUniFrac, vegan Suggests R.rsp VignetteBuilder R.rsp Author Yi-Juan Hu and Glen A. Satten Maintainer Yi-Juan Hu Description The LDM package provides a single analysis path that includes distance-based ordination, global tests of any effect of the microbiome, and tests of the effects of individual OTUs (i.e., operational taxonomic units) with false discovery rate (FDR)-based correction for multiple testing. It accommodates both continuous and discrete variables (e.g., clinical outcomes, environmental factors, treatment groups) as well as interaction terms to be tested either singly or in combination, allows for adjustment of confounding covariates, and uses permutation-based p-values that can control for correlation (e.g., repeated measurements on the same individual). It can also be applied to transformed, and an `omnibus' test can easily combine results from analyses conducted on different transformation scales. License GPL (>=2) NeedsCompilation no RoxygenNote R topics documented: adjust..by.covariates ldm ortho.decomp permanovac test.ldm.fromfile throat.meta throat.otu.tab throat.tree Index 16 1

2 2 adjust..by.covariates adjust..by.covariates Adjust by covariates Description This function produces adjusted distance matrix and otu table (if provided) after removing the effects of covariates (e.g., confounders). Observations with any missing are removed. Usage adjust..by.covariates(formula = NULL, = NULL, otu.table = NULL, tree = NULL, dist.method = "bray", dist = NULL, square.dist = TRUE, center.dist = TRUE, scale.otu.table = TRUE, center.otu.table = TRUE) Arguments formula a symbolic description of the covariate model with the form ~ model, where model is specified in the same way as for lm or glm. For example, ~ a + b specifies a model with the main effects of covariates a and b, and ~ a*b, equivalently ~ a + b + a:b, specifies a model with the main effects of a and b as well as their interaction. otu.table tree dist.method dist square.dist an optional frame, list or environment (or object coercible by as..frame to a frame) containing the covariates. If not found in, the covariates are taken from environment(formula), typically the environment from which adjust..by.covariates is called. The default is NULL. the n.obs by n.otu matrix of read counts. If provided, the adjusted (and columncentered) otu.table at both the frequency scale and arcsin-root-transformed frequency scale are outputted. If provided, it is also used for calculating the distance matrix unless the distance matrix is directly imported through dist. The default is NULL. a phylogenetic tree. Used for calculating a phylogenetic-tree-based distance matrix. Not needed if the calculation of requested distance does not require a phylogenetic tree, or if the distance matrix is directly imported through dist. The default is NULL. method for calculating the distance measure, partial match to all methods supported by vegdist in the vegan package (i.e., "manhattan", "euclidean", "canberra", "bray", "kulczynski", "jaccard", "gower", "altgower", "morisita", "horn", "mountford", "raup", "binomial", "chao", "cao", "mahalanobis") as well as "hellinger" and "wt-unifrac". The default is "bray". For more details, see the dist.method argument in the ldm function. a distance matrix. It can be the original distance matrix and then squared/centered by the default square.dist=true/center.dist=true; it can also be squared/centered a priori and used with square.dist=false/center.dist=false. If not provided, the distance matrix is calculated using otu.table, tree, and dist.method. The default is NULL. a logical variable indicating whether to square the distance matrix. The default is TRUE.

3 ldm 3 Value center.dist a logical variable indicating whether to center the distance matrix as described by Gower (1966). The default is TRUE. scale.otu.table a logical variable indicating whether to scale (i.e., divide) the rows of the OTU table by the library size, so the read counts becomes relative frequencies. The default is TRUE. It can be set to FALSE if the input otu.table has been scaled a priori. center.otu.table a logical variable indicating whether to center the columns of the OTU table. The OTU table should be centered if the distance matrix has been centered. The default is TRUE. a list consisting of adj.dist x.freq x.tran the (squared/centered) distance matrix after adjustment of covariates. the (column-centered) frequency-scale matrix after adjustment of covariates. the (column-centered) arcsin-root-transformed matrix after adjustment of covariates. Author(s) Yi-Juan Hu <yijuan.hu@emory.edu>, Glen A. Satten <gas0@cdc.gov> Examples adj. <- adjust..by.covariates(formula= ~ Sex + AntibioticUse, =throat.meta, otu.table=throat.otu.tab, dist.method="bray") # # Use the adjusted distance matrix for ordination # PCs <- eigen(adj.$adj.dist, symmetric=true) color = rep("blue", length(smokingstatus)) w = which(smokingstatus=="smoker") color[w] = "red" plot(pcs$vectors[,1], PCs$vectors[,2], xlab="pc1", ylab="pc2", col=color, main="smokers vs. non-smokers") legend(x="topleft", legend=c("smokers","non-smokers"), pch=c(21,21), col=c("red","blue"), lty=0) ldm Testing hypotheses using a linear decomposition model (LDM) Description This function allows you to simultaneously test the global association with the overall microbiome composition and individual OTU associations to give coherent results. It is capable of handling complex design features such as confounders, interactions, and clustered.

4 4 ldm Usage ldm(formula, =.GlobalEnv, tree = NULL, dist.method = "bray", dist = NULL, cluster.id = NULL, permute.within.clusters = FALSE, permute.between.clusters = FALSE, n.perm.max = 5000, n.rej.stop = 10, seed = NULL, test.global = TRUE, test.otu = TRUE, fdr.nominal = 0.1, outfile.prefix = NULL, decomposition = "orthogonal", square.dist = TRUE, center.dist = TRUE, scale.otu.table = TRUE, center.otu.table = TRUE) Arguments formula tree dist.method dist a symbolic description of the model to be fitted. The details of model specification are given under "Details". an optional frame, list or environment (or object coercible by as..frame to a frame) containing the covariates of interest and confounding covariates. If not found in, the covariates are taken from environment(formula), typically the environment from which ldm is called. The default is.globalenv. a phylogenetic tree. Used for calculating a phylogenetic-tree-based distance matrix. Not needed if the calculation of the requested distance does not involve a phylogenetic tree, or if the distance matrix is directly imported through dist. method for calculating the distance measure, partial match to all methods supported by vegdist in the vegan package (i.e., "manhattan", "euclidean", "canberra", "bray", "kulczynski", "jaccard", "gower", "altgower", "morisita", "horn", "mountford", "raup", "binomial", "chao", "cao", "mahalanobis") as well as "hellinger" and "wt-unifrac". Note that the OTU table will be converted to the frequency scale and binary is set to FALSE (i.e., not allowing calculation of presence-absence distances) when we internally call vegdist for its supported distances. The Hellinger distance measure (dist.method="hellinger") takes the form 0.5*E, where E is the Euclidean distance between the square-roottransformed frequency. The weighted UniFrac distance (dist.method="wt-unifrac") is calculated by interally calling GUniFrac in the GUniFrac package. Not used when anything other than dist=null is specified for dist. The default is "bray". a distance matrix. It can be the original distance matrix and then squared/centered by the default square.dist=true/center.dist=true; it can also be squared/centered a priori and used with square.dist=false/center.dist=false. If not specified, the distance matrix is calculated using otu.table, tree, and dist.method. The default is NULL. cluster.id cluster identifiers. Unequal cluster size is allowed. The default is NULL. Use the default if the observations are not in clusters (i.e., independent). permute.within.clusters a logical value indicating whether to permute within clusters. The default is FALSE. permute.between.clusters a logical value indicating whether to permute between clusters. The default is FALSE. n.perm.max the maximum number of permutations to run. The default is n.rej.stop seed the minimum number of rejections (i.e., the permutation statistic exceeds the observed statistic) to obtain before stopping. The default is 10. an integer seed for the random number generator in the permutation procedure. The default is NULL; with the default value, an integer seed will be generated

5 ldm 5 test.global test.otu internally and randomly. In either case, the integer seed will be stored in the output object (or, if outfile.prefix is specified, the integer seed will be stored in the output file "outfile.prefix_seed.txt") in case the user wants to reproduce the permutations. a logical value indicating whether to perform the global test. The default is TRUE. a logical value indicating whether to perform the OTU-specific tests. The default is TRUE. fdr.nominal the nominal FDR value. The default is 0.1. outfile.prefix a character string specifying the prefix (including the path) of the external files that are used to save intermediate statistics. If this is specified, the intermediate statistics are saved in the external files and the test results are not generated from the ldm function; the user should use the test.ldm.fromfile function to summarize the intermediate statistics and generate test results. This feature can be used to parallelize the permutation onto multiple machines, each starting with a different seed. The default is NULL, under which the intermediate statistics are not saved externally and the test results are directly generated from the ldm function. decomposition square.dist a character string that takes either the default value "orthogonal" corresponding to the LDM method or "non-orthogonal" corresponding to the Redundancy Analysis (RA) or partial RA if more than one set of variables is specified. a logical variable indicating whether to square the distance matrix. The default is TRUE. center.dist a logical variable indicating whether to center the distance matrix as described by Gower (1966). The default is TRUE. scale.otu.table a logical variable indicating whether to scale the rows of the OTU table (i.e., divide by the library size), so the read counts becomes relative frequencies. The default is TRUE. It can be set to FALSE if the input otu.table has been scaled a priori. center.otu.table a logical variable indicating whether to center the columns of the OTU table. The OTU table should be centered if the distance matrix has been centered. The default is TRUE. Details The formula has the form otu.table ~ (first set of covariates) + (second set of covariates)... + (last set of covariates) or otu.table confounders ~ (first set of covariates) + (second set of covariates)... + (last set of covariates) where otu.table is the OTU table with rows for samples and columns for OTUs and each set of covariates are enclosed in parentheses. Each set of covariates is tested individually and sequentially (i.e., after accounting for the previous sets of covariates). Suppose there is an otu table y and a frame meta that contains 4 covariates, a, b, c and d. Some valid formulas would be (along with =meta): y ~ a + b + c + d ### no confounders, 4 models (i.e., sets of covariates)

6 6 ldm y ~ (a+b) + (c+d) ### no confounders, 2 models each having 2 covariates; it is Okay even when a has values A and B while b has numeric values y b ~ (a+c) + d ### b is a confounder, model 1 is (a+c), and model 2 is d y b+c ~ a*d ### there are 2 confounders b and c; there is 1 model for the three terms a, d, and a:d (interaction). It is equivalent to y b+c ~ (a+d+a:d). y as.factor(b) ~ (a+d) + a:d ### now confounder b will be treated as a factor variable, model 1 will have the main effects a and d, and model 2 will have only the interaction between a and d y as.factor(b) ~ (a) + (d) + (a:d) ### there are 3 models a, d, and a:d. It is Okay to put paratheses around a single variable. LDM uses two sequential stopping criteria. For the global test, LDM uses the stopping rule of Besag and Clifford (1991), which stops permutation when a pre-specified minimum number (default=10) of rejections (i.e., the permutation statistic exceeded the observed test statistic) has been reached. For the OTU-specific tests, LDM uses the stopping rule of Sandve et al. (2011), which stops permutation when every OTU test has either reached the pre-specified number (default=10) of rejections or yielded a q-value that is below the nominal FDR level (default=0.1). As a convention, we call a test "stopped" if the corresponding stopping criterion has been satisfied. Although all tests are always terminated if a pre-specified maximum number (default=5000) of permutations have been generated, some tests may not have "stopped" and hence their results have not achieved good precision. Value a list consisting of b the matrix B as defined in Hu and Satten (2018) b.use.freq b.use.tran adj.dist x.freq d.freq v.freq x.tran the columns of b used for orthogonal decomposition of x.freq. When the number of columns of b exceeds the rank of x.freq, we use only a subset of the columns of Br, selecting those that have the largest eigenvalue (in absolute value) in the decomposition of the residual distance matrix that are not orthogonal to the column space of x.freq. the columns of b used for orthogonal decomposition of x.tran. When the number of columns of b exceeds the rank of x.tran, we use only a subset of the columns of Br, selecting those that have the largest eigenvalue (in absolute value) in the decomposition of the residual distance matrix that are not orthogonal to the column space of x.tran. the (squared/centered) distance matrix after adjusting for confounders. If there are no confounders, it is simply the (squared/centered) distance matrix. the (column-centered) frequency-scale matrix after adjusting for confounders (if there are any) a vector of the nonnegative diagonal elements of D that minimizes x.freq - b D v^t ^2 subject to v^tv=i in orthogonal composition or satisfies b^t x.freq = D v^t in non-orthogonal decomposition the v matrix with unit columns that minimizes x.freq - b D v^t ^2 subject to v^tv=i in orthogonal composition or satisfies b^t x.freq = D v^t in non-orthogonal decomposition the (column-centered) arcsin-root-transformed matrix after adjusting for confounders (if there are any).

7 ldm 7 d.tran v.tran a vector of the nonnegative diagonal elements of D that minimizes x.tran - b D v^t ^2 subject to v^tv=i in orthogonal composition or satisfies b^t x.tran = D v^t in non-orthogonal decomposition the v matrix with unit columns that minimizes x.tran - b D v^t ^2 subject to v^tv=i in orthogonal composition or satisfies b^t x.tran = D v^t in non-orthogonal decomposition VE.global.freq variance explained by each set of covariates, based on the frequency-scale VE.global.tran variance explained by each set of covariates, based on the arcsine-root-transformed frequency VE.otu.freq VE.otu.tran p.global.freq p.global.tran p.global.omni p.otu.freq p.otu.tran p.otu.omni q.otu.freq q.otu.tran contributions from each OTU to VE.global.freq contributions from each OTU to VE.global.tran p-values for the global test of each set of covariates based on the frequency-scale p-values for the global test of each set of covariates based on the arcsin-roottransformed frequency p-values for the global test of each set of covariates based on the omnibus statistics, which are the minima of the p-values obtained from the frequency scale and the arcsin-root-transformed frequency as the final test statistics, and use the corresponding minima from the permuted to simulate the null distributions p-values for the OTU-specific tests based on the frequency scale p-values for the OTU-specific tests based on the arcsin-root-transformed frequency p-values for the OTU-specific tests based on the omnibus statistics q-values (i.e., FDR-adjusted p-values) for the OTU-specific tests based on the frequency scale q-values for the OTU-specific tests based on the arcsin-root-transformed frequency q.otu.omni q-values for the OTU-specific tests based on the omnibus statistics n.perm.completed number of permutations completed global.tests.stopped a logical value indicating whether the stopping criterion has been met by all global tests otu.tests.stopped a logical value indicating whether the stopping criterion has been met by all OTU-specific tests seed a single-value integer seed that is user supplied or internally generated Author(s) Yi-Juan Hu <yijuan.hu@emory.edu>, Glen A. Satten <gas0@cdc.gov> References Hu YJ, Satten GA. (2018) Testing hypotheses about microbiome using an ordination-based linear decomposition model. biorxiv:doi.org/ /

8 8 ortho.decomp Examples # # fit only # fit <- ldm(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", n.perm.max=0) # # test the global hypothese only # res1.ldm <- ldm(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", test.global=true, test.otu=false, n.perm.max=10000, seed=1) # # test both the global hypothese and individual OTUs # res2.ldm <- ldm(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", test.global=true, test.otu=true, fdr.nominal=0.1, n.perm.max=85600, seed=100) # # test both the global hypothese and individual OTUs; # save the intermediate statistis to external files; # parallelize permutation # dir.create("tmp_files") # create the directory for external files to store intermediate statistics seed.seq=c(81426, 92877, 14748, 74980) # generated by sample.int(100000, size=4, replace=false) res3.tmp <- ldm(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", test.global=true, test.otu=true, outfile.prefix="tmp_files/smk_age", n.perm.max=10000, seed=seed.seq[1]) # calling ldm repeatedly with seed=seed.seq[2], seed.seq[3], seed.seq[4] res3.ldm <- test.ldm.fromfile(infile.prefix="tmp_files/smk_age", n.perm.available=40000, test.global=true, test.otu=true, fdr.nominal=0.1) # # clustered # res4.ldm <- ldm(formula=sim.otu.tab (C1+C2) ~ Y, =sim.meta,dist.method="bray", cluster.id=id, permute.between.clusters=true, test.global=true, test.otu=true, fdr.nominal=0.1, n.perm.max=5000, seed=1) ortho.decomp Orthogonal decomposition

9 ortho.decomp 9 Description Usage This function performs the (approximate) orthogonal decomposition which gives a duality between principle components of a matrix x and linear combinations of a set of orthogonal directions b (e.g., principle components of a distance matrix). Specifically, it finds the matrix v with orthonormal columns (i.e.,v^t v=i) and the diagonal matrix D with nonnegative diagonal entries that minimize x - b D v^t ^2. ortho.decomp(x, b, x.svd = NULL, tol.conv = 10^-8, n.iter.max = 1000, verbose = FALSE) Arguments x b Value x.svd tol.conv the n.obs by n.otu matrix the n.obs by n.vec matrix with orthonormal columns, e.g., the matrix having columns given by all eigenvectors of a distance matrix the result object of svd(x). The default is NULL and then svd(x) is performed internally. the stopping criterion for the tandem algorithm. The algorithm is stopped when the largest change in any element of d or v is smaller than tol.conv. The default is 10^-8. n.iter.max the maximum number of iterations to take before stopping. The default is verbose a list consisting of a logical variable indicating whether to return detailed information on the decomposition. The default is FALSE. v d n.iter fail b.hat r2 converge an n.otu by n.vec matrix having orthonormal columns that give the directions (OTU loading) in feature space a vector of length n.vec with the diagonal elements of D the number of iterations used to converge equals 0 if the algorithm converged to the desired tolerance or 1 if not an n.obs by n.vec matrix containing the estimated values of b. Returned when verbose=true. R2 between columns of b and columns of b.hat. Returned when verbose=true. A message informing whether the algorithm converged or not and the number of iterations used. Returned when verbose=true. Author(s) Yi-Juan Hu <yijuan.hu@emory.edu>, Glen A. Satten <gas0@cdc.gov> References Satten G.A., Tyx R., Rivera A.J., Stanfill S. (2017). Restoring the Duality between Principal Components of a Distance Matrix and Linear Combinations of Predictors, with Application to Studies of the Microbiome. PLoS ONE 12(1):e doi: /journal.pone

10 10 permanovac Examples res <- ortho.decomp( x=.matrix, b=dist.eigen.vectors, verbose=true ) permanovac PERMANOVA test of association Description This function performs the PERMANOVA test that can allow adjustment of confounders and control of clustered. As in ldm, permanovac allows multiple sets of covariates to be tested, in the way that the sets are entered sequentially and the variance explained by each set is that part that remains after the previous sets have been fit. Usage permanovac(formula, =.GlobalEnv, tree = NULL, dist.method = "bray", cluster.id = NULL, permute.within.clusters = FALSE, permute.between.clusters = FALSE, n.perm.max = 5000, n.rej.stop = 10, seed = NULL, square.dist = TRUE, center.dist = TRUE) Arguments formula tree dist.method cluster.id a symbolic description of the model to be fitted in the form of.matrix ~ sets of covariates or.matrix confounders ~ sets of covariates. The details of model specification are given in "Details" of ldm. The only difference between the model here and that for ldm is that, the.matrix here is either an OTU table or a distance matrix. If it is an OTU table, the distance matrix will be calculated internally using the OTU table, tree, and dist.method. If it is a distance matrix, it can be the original distance matrix and then squared/centered by the default square.dist=true/center.dist=true; it can also be squared/centered a priori and used with square.dist=false/center.dist=false. an optional frame, list or environment (or object coercible by as..frame to a frame) containing the covariates of interest and confounding covariates. If not found in, the covariates are taken from environment(formula), typically the environment from which permanovac is called. The default is.globalenv. a phylogenetic tree. Used for calculating a phylogenetic-tree-based distance matrix. Not needed if the calculation of the requested distance does not involve a phylogenetic tree, or if a distance matrix is directly imported through formula. method for calculating the distance measure, partial match to all methods supported by vegdist in the vegan package (i.e., "manhattan", "euclidean", "canberra", "bray", "kulczynski", "jaccard", "gower", "altgower", "morisita", "horn", "mountford", "raup", "binomial", "chao", "cao", "mahalanobis") as well as "hellinger" and "wt-unifrac". Not used if a distance matrix is specified in formula. The default is "bray". For more details, see the dist.method argument in the ldm function. cluster identifiers. Unequal cluster size is allowed. The default is NULL. Use the default if the observations are not in clusters (i.e., independent).

11 test.ldm.fromfile 11 permute.within.clusters a logical value indicating whether to permute within clusters. The default is FALSE. permute.between.clusters a logical value indicating whether to permute between clusters. The default is FALSE. n.perm.max the maximum number of permutations to run. The default is n.rej.stop seed square.dist center.dist the minimum number of rejections (i.e., the permutation statistic exceeds the observed statistic) to obtain before stopping. The default is 10. an integer seed for the random number generator in the permutation procedure. The default is NULL; with the default value, an integer seed will be generated internally and randomly. In either case, the integer seed will be stored in the output object. a logical variable indicating whether to square the distance matrix. The default is TRUE. a logical variable indicating whether to center the distance matrix as described by Gower (1966). The default is TRUE. Value a list consisting of F.statistics p.permanova F statistics for the global test of each set of covariates p-values for the global test of each set of covariates n.perm.completed number of permutations completed permanova.stopped a logical value indicating whether the stopping criterion has been met by all global tests seed a single-value integer seed that is user supplied or internally generated Author(s) Yi-Juan Hu <yijuan.hu@emory.edu>, Glen A. Satten <gas0@cdc.gov> Examples res.permanova <- permanovac(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", seed=1) test.ldm.fromfile Testing hypotheses based on saved intermediate statistics Description This function summarizes the intermediate statistics saved in external files into p-values and q- values.

12 12 test.ldm.fromfile Usage test.ldm.fromfile(infile.prefix = NULL, n.perm.available, n.rej.stop = 10, test.global = TRUE, test.otu = TRUE, fdr.nominal = 0.1) Arguments Value infile.prefix a character string specifying the prefix (including the path) of the external files that save the intermediate statistics n.perm.available an (approximate) estimate of the total number of permutations in the directory specified by infile.prefix. Used for managing memory when reading large files. n.rej.stop test.global test.otu the minimum number of rejections (i.e., the permutation statistic exceeds the observed statistic) to obtain before stopping. The default is 10. a logical value indicating whether to perform the global test. The default is TRUE. a logical value indicating whether to perform the OTU-specific tests. The default is TRUE. fdr.nominal the nominal FDR value. The default is 0.1. a list consisting of p.global.freq p.global.tran p.global.omni p.otu.freq p.otu.tran p.otu.omni q.otu.freq q.otu.tran p-values for the global test of each set of covariates based on the frequency scale p-values for the global test of each set of covariates based on the arcsin-roottransformed frequency p-values for the global test of each set of covariates based on the omnibus statistics, which are the minima of the p-values obtained from the frequency scale and the arcsin-root-transformed frequency as the final test statistics, and use the corresponding minima from the permuted to simulate the null distributions p-values for the OTU-specific tests based on the frequency scale p-values for the OTU-specific tests based on the arcsin-root-transformed frequency p-values for the OTU-specific tests based on the omnibus statistics q-values (i.e., FDR-adjusted p-values) for the OTU-specific tests based on the frequency scale q-values for the OTU-specific tests based on the arcsin-root-transformed frequency q.otu.omni q-values for the OTU-specific tests based on the omnibus statistics n.perm.completed number of permutations completed global.tests.stopped a logical value indicating whether the stopping criterion has been met by all global tests otu.tests.stopped a logical value indicating whether the stopping criterion has been met by all OTU-specific tests

13 throat.meta 13 Author(s) Yi-Juan Hu Glen A. Satten Examples # # test both the global hypothese and individual OTUs; # save the intermediate statistis to external files; # parallelize permutation # dir.create("tmp_files") # create the directory seed.seq=c(81426, 92877, 14748, 74980) # generated by sample.int(100000, size=4, replace=false) res3.tmp <- ldm(formula=throat.otu.tab (Sex+AntibioticUse) ~ (SmokingStatus+PackYears)+Age, =throat.meta, dist.method="bray", test.global=true, test.otu=true, outfile.prefix="tmp_files/smk_age", n.perm.max=10000, seed=seed.seq[1]) # calling ldm repeatedly with seed=seed.seq[2], seed.seq[3], seed.seq[4] res3.ldm <- test.ldm.fromfile(infile.prefix="tmp_files/smk_age", n.perm.available=40000, test.global=true, test.otu=true, fdr.nominal=0.1) throat.meta Meta of the throat microbiome samples Description Usage Format Source This set includes samples from the microbiome of the nasopharynx and oropharynx on each side of the body. It were generated to study the effect of smoking on the microbiota of the upper respiratory tract in 60 individuals, 28 smokers and 32 nonsmokers. ("throat.meta") A frame with 60 observations on 16 variables. Charlson ES, Chen J, Custers-Allen R, Bittinger K, Li H, et al. (2010) Disordered Microbial Communities in the Upper Respiratory Tract of Cigarette Smokers. PLoS ONE 5(12): e References R package "GUniFrac" Examples (throat.meta)

14 14 throat.tree throat.otu.tab OTU count table from 16S sequencing of the throat microbiome samples Description This set contains 60 subjects with 28 smokers and 32 nonsmokers. Microbiome were collected from right and left nasopharynx and oropharynx region to form an OTU table with 856 OTUs. Usage ("throat.otu.tab") Format A frame with 60 observations on 856 variables. Source Charlson ES, Chen J, Custers-Allen R, Bittinger K, Li H, et al. (2010) Disordered Microbial Communities in the Upper Respiratory Tract of Cigarette Smokers. PLoS ONE 5(12): e References R package "GUniFrac" Examples (throat.otu.tab) throat.tree UPGMA tree of the OTUs from 16S sequencing of the throat microbiome samples Description The OTU tree is constructed using UPGMA on the K80 distance matrix of the OTUs. It is a rooted tree of class "phylo". Usage ("throat.tree") Format List of 4 frames.

15 throat.tree 15 Source Charlson ES, Chen J, Custers-Allen R, Bittinger K, Li H, et al. (2010) Disordered Microbial Communities in the Upper Respiratory Tract of Cigarette Smokers. PLoS ONE 5(12): e References R package "GUniFrac" Examples (throat.tree)

16 Index Topic PCA adjust..by.covariates, 2 ortho.decomp, 8 Topic sets throat.meta, 13 throat.otu.tab, 14 throat.tree, 14 Topic distance adjust..by.covariates, 2 ortho.decomp, 8 Topic microbiome adjust..by.covariates, 2 ldm, 3 ortho.decomp, 8 permanovac, 10 test.ldm.fromfile, 11 Topic ordination adjust..by.covariates, 2 ortho.decomp, 8 adjust..by.covariates, 2 ldm, 3 ortho.decomp, 8 permanovac, 10 test.ldm.fromfile, 11 throat.meta, 13 throat.otu.tab, 14 throat.tree, 14 16