comp_by_deleting_random_knowns_t_v3: Create bootstrap networks and compare network measures for...
In ConesaLab/MDM: Complete MDM Network Analysis for 16S rRNA Data

Description Usage Arguments Value Examples

Bootstrap networks are generated, removing an equal number of random knowns as MDM at each rank and all network measures are recalculated for nodes in these networks.To compare network measures for Original, Without Unknown, and Bootstrap networks for all taxonomic rank level, a dataframe is created, showing the node values for degree, betweenness, and closeness for nodes present in each network type at each taxonomic rank.

1 2	comp_by_deleting_random_knowns_t_v3(orig_graph, new_wo_unk_graph, orig_phylo, orig_df, degree_df_wo_unk, iter = 100, mc.coreval=2)

`orig_graph`	Network including all taxa (Original network), produced from get_back_res_meeting_min_occ function
`new_wo_unk_graph`	Network excluding MDM taxa (Without Unknown network), produced by get_graph_wo_unk function
`orig_phylo`	Phyloseq including all taxa meeting threshold, produced from get_back_res_meeting_min_occ function
`orig_df`	Dataframe of degree, betweenness, and closeness scores for all nodes in Original network, produced by degree_calc_df function
`degree_df_wo_unk`	Dataframe of degree, betweenness, and closeness scores for all nodes in Without Unknown Network, produced by get_degree_df_wo_unk function
`iter`	Number of iterations of random subsampling and removal of knowns. Default is 100 iterations.
`mc.coreval`	Number of cores to use for this function to parallelize this long process. Default is 2 to allow this function to run on a computer.

Returns dataframe of 4 columns, respectively indicating taxonomic rank (Kingdom to Species), network type (Original, Bootstrap, or Without_unk), measure (degree, betweenness, closeness), and score for each node.

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (orig_graph, new_wo_unk_graph, orig_phylo, orig_df,
    degree_df_wo_unk, iter = 100, mc.coreval=2)
{
    orig_vertices <- as.vector(names(V(orig_graph)))
    measurements <- c("degree", "bw", "closeness")
    all_net_tax_level_plot_l <- do.call(rbind, parallel::mclapply(seq_along(new_wo_unk_graph),
        mc.cores = mc.coreval, function(i) {
            rank_level = names(new_wo_unk_graph)[[i]]
            print(paste0("Rank level: ", rank_level))
            num_nodes_to_remove <- length(V(orig_graph)) - length(V(new_wo_unk_graph[[rank_level]]))
            unk_names <- setdiff(names(V(orig_graph)), names(V(new_wo_unk_graph[[rank_level]])))
            orig_vertices_wo_unk <- intersect(names(V(orig_graph)),
                names(V(new_wo_unk_graph[[rank_level]])))
            bstrap_df_l <- lapply(seq_len(iter), function(i) {
                vecnames <- sample(x = orig_vertices_wo_unk,
                  size = num_nodes_to_remove, replace = TRUE)
                wo_unk_graph <- igraph::delete_vertices(orig_graph,
                  which(names(V(orig_graph)) %in% vecnames))
                boot_degree_df <- degree_calc_f(wo_unk_graph)
                if (!identical(length(rownames(boot_degree_df)),
                  length(rownames(degree_df_wo_unk[[rank_level]])))) {
                  boot_degree_df = boot_degree_df[seq_along(rownames(degree_df_wo_unk[[rank_level]])),
                    ]
                }
                boot_degree_df[, measurements]
            })
            do.call(rbind, lapply(measurements, function(act_meas) {
                orig_meas <- orig_df[[act_meas]]
                wo_unk_meas <- degree_df_wo_unk[[rank_level]][[act_meas]]
                bstrap_meas <- rowMeans(do.call(cbind, lapply(bstrap_df_l,
                  function(x) x[, act_meas])))
                data.frame(rank_level = rank_level, type = factor(c(rep("Original",
                  length(orig_meas)), rep("Without_unknown",
                  length(wo_unk_meas)), rep("Bootstrap", length(bstrap_meas))),
                  levels = c("Original", "Without_unknown", "Bootstrap")),
                  measure = act_meas, data = c(orig_meas, wo_unk_meas,
                    bstrap_meas))
            }))
        }))
    return(all_net_tax_level_plot_l)
  }