{ "cells": [ { "cell_type": "code", "execution_count": null, "id": "prostate-definition", "metadata": {}, "outputs": [], "source": [ "############# Taxonomic profiling\n", "\n", "############# Merge using BBMap\n", "\n", "$ bbmerge-auto.sh in1=16S-Sample-01_R1.fastq in2=16S-Sample-01_R2.fastq out=16S-Sample-01.merged.fastq outu=16S-Sample-01.unmerged.fastq rsem extend2=50 k=62 -Xmx3613m \n", "\n", "$ gzip 16S-Sample-01.merged.fastq\n", "\n", "############# Filter for quality using Trimmomatic\n", "\n", "$ TrimmomaticSE -threads 35 -phred33 16S-Sample-01.merged.fastq.gz 16S-Sample-01.trimmed.fastq.gz LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15\n", "\n", "$ gunzip 16S-Sample-01.trimmed.fastq.gz\n", " \n", "############# Chimeras were removed using USEARCH against the SILVA database\n", "\n", "$ usearch9.2.64_i86linux64 -uchime2_ref 16S-Sample-01.trimmed.fastq -db SILVA_SSU.fasta -notmatched 16S-Sample-01.fastq -strand plus -mode high_confidence\n", "\n", "$ gzip 16S-Sample-01.fastq\n", "\n", "############# Phylogenetic assignment and relative quantification\n", "\n", "$ kraken2 --db oral_microbiome_UNITE/ --threads 40 --output 16S-Sample-01.kraken2 --report 16S-Sample-01.kreport2 --gzip-compressed 16S-Sample-01.fastq.gz\n", "\n", "$ bracken -d oral_microbiome_UNITE/ -t 40 -i 16S-Sample-01.kreport2 -o 16S-Sample-01.bracken -r 250 -l S \n", "\n", "$ kreport2mpa.py -r 16S-Sample-01_bracken_species.kreport2 -o 16S-Sample-01.mpa ### Run these for each sample. Generates a mpa file from bracken reports.\n", "\n", "### Merge the results (.mpa files) of all samples into a table:\n", "\n", "$ ./merge_metaphlan_tables.py 16S-Sample-01.mpa 16S-Sample-02.mpa … > merged_OTUs_table.tab\n", "\n", "############# Custom database of the ITS sequences with selected fungi species that had been previously identified in oral samples and the ITS sequences of Entamoeba gingivalis. \n", "\n", "\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "id": "close-honduras", "metadata": {}, "outputs": [], "source": [ "############# General procedure used to generate a phyloseq object\n", "############# OTUS.csv was obtained from 'merged_OTUs_table.tab' table selecting only for species.This is the table that was used throughout all the analyses described below.\n", "### In R\n", "\n", "library(phyloseq)\n", "\n", "otus<-read.csv('OTUS.csv', header = TRUE, row.names = 1) ## Count table\n", "otus<-t(otus)\n", "taxmat<-read.csv(\"taxmat.csv\", header = TRUE, row.names = 1) ## Taxonomy table\n", "samples<-read.csv(\"Sample_data.csv\", header = TRUE, row.names =1) ## Sample information table\n", "taxmat<-as.matrix(taxmat)\n", "otus<-as.matrix(otus)\n", "\n", "OTU = otu_table(otus, taxa_are_rows = FALSE)\n", "TAX = tax_table(taxmat)\n", "SAM = sample_data(samples)\n", "\n", "physeq = phyloseq(OTU, TAX, SAM) ### generate phyloseq object\n", "\n", "random_tree = rtree(ntaxa(physeq), rooted=TRUE, tip.label=taxa_names(physeq))\n", "\n", "physeq = phyloseq(OTU, TAX, SAM, random_tree) ## Phyloseq object to be used in analysis below\n", "\n", "sample_data(physeq)$group = as.character(sample_data(physeq)$group)\n", "get_variable(physeq,\"group\")\n" ] }, { "cell_type": "code", "execution_count": null, "id": "looking-lafayette", "metadata": {}, "outputs": [], "source": [ "############# Figure 1a. Barplot representation of temporal changes in frequencies of species in the three groups studied\n", "### In R\n", "\n", "library(phyloseq)\n", "library(ape)\n", "library(dplyr)\n", "library(ggplot2)\n", "library(microbiomeSeq)\n", "\n", "otus<-read.csv('OTUS.csv', header = TRUE, row.names = 1)\n", "otus=otus+1\n", "otus<-t(otus)\n", "taxmat<-read.csv(\"taxmat.csv\", header = TRUE, row.names = 1)\n", "samples<-read.csv(\"Sample_data.csv\", header = TRUE, row.names =1)\n", "\n", "taxmat<-as.matrix(taxmat)\n", "otus<-as.matrix(otus)\n", "\n", "OTU = otu_table(otus, taxa_are_rows = FALSE)\n", "TAX = tax_table(taxmat)\n", "SAM = sample_data(samples)\n", "\n", "physeq = phyloseq(OTU, TAX, SAM) ### generate phyloseq object\n", "random_tree = rtree(ntaxa(physeq), rooted=TRUE, tip.label=taxa_names(physeq))\n", "physeq = phyloseq(OTU, TAX, SAM, random_tree)\n", "sample_data(physeq)$group = as.character(sample_data(physeq)$group)\n", "get_variable(physeq,\"group\")\n", "\n", "### Normalize number of reads in each sample \n", "\n", "physeq <- normalise_data(physeq, norm.method = \"edgernorm\")\n", "\n", "############# Time series barplot \n", "\n", "erie_phylum <- physeq %>%\n", " tax_glom(taxrank = \"Species\") %>% # agglomerate at species level\n", " transform_sample_counts(function(x) {x/sum(x)} ) %>% # Transform to rel. abundance\n", " psmelt() %>% # Melt to long format\n", " filter(Abundance > 0.1) %>% # Filter out low abundance taxa\n", " arrange(Species) \n", "\n", "phylum_colors <-c(\"maroon1\",\"cornsilk2\",\"magenta1\",\"darkgoldenrod1\",\"lightyellow1\",\"antiquewhite2\",\"yellow1\",\"deeppink1\",\"lavenderblush1\",\"mediumorchid2\",\"lightblue3\",\"springgreen1\",\"aquamarine2\",\"chartreuse3\",\"turquoise4\",\"bisque1\",\"palevioletred1\",\"deepskyblue4\",\"turquoise1\",\"lightblue2\",\"khaki1\",\"cornsilk1\",\"lightskyblue3\",\"magenta2\",\"lightcyan3\",\"deeppink2\",\"yellow3\",\"indianred2\",\"darkseagreen3\",\"lightyellow4\",\"ivory2\",\"olivedrab1\",\"ivory1\",\"cyan1\",\"tan1\",\"coral1\",\"darkorchid1\",\"mediumpurple4\",\"magenta3\",\"honeydew1\",\"gold1\",\"lightsteelblue1\",\"lightsalmon3\",\"indianred4\",\"lightsalmon4\",\"deepskyblue2\",\"antiquewhite3\",\"darkorange1\",\"springgreen2\",\"gold2\",\"orange4\",\"lemonchiffon1\",\"maroon4\",\"ivory4\",\"mediumpurple1\",\"maroon3\",\"chocolate2\",\"purple3\",\"honeydew4\",\"plum2\",\"aquamarine3\",\"darkorchid3\",\"tan3\",\"lemonchiffon3\",\"lemonchiffon4\",\"darkseagreen2\",\"chocolate1\",\"mediumorchid3\",\"seegreen1\",\"brown2\",\"hotpink1\")\n", "\n", " ggplot(erie_phylum, aes(x = time, y = Abundance, fill = Species)) + \n", " facet_grid(group~.) +\n", " geom_bar(stat = \"identity\") +\n", " scale_fill_manual(values = phylum_colors) +\n", " # Remove x axis title\n", " theme(axis.title.x = element_blank()) + \n", " #\n", " guides(fill = guide_legend(reverse = TRUE, keywidth = 1, keyheight = 1)) +\n", " ylab(\"Relative Abundance (Species > 5%) \\n\") +\n", " ggtitle(\"Species Composition\") \n" ] }, { "cell_type": "code", "execution_count": null, "id": "motivated-accordance", "metadata": {}, "outputs": [], "source": [ "############# Figure 1b. Violin plot of Shannon alpha-diversity. Using the phyloseq object ‘physeq’\n", "### In R\n", "\n", "ps1 <- prune_taxa(taxa_sums(physeq) > 0, physeq)\n", "tab <- microbiome::alpha(ps1, index = \"all\")\n", "ps1.meta <- meta(ps1)\n", "ps1.meta$Shannon <- tab$diversity_shannon \n", "ps1.meta$InverseSimpson <- tab$diversity_inverse_simpson\n", "\n", "# create a list of pairwise comaprisons\n", "ps1.meta$group<-factor(ps1.meta$group)\n", "bmi <- levels(ps1.meta$group) # get the variables\n", "# make a pairwise list that we want to compare.\n", "bmi.pairs <- combn(seq_along(bmi), 2, simplify = FALSE, FUN = function(i)bmi[i])\n", "\n", "print(bmi.pairs)\n", "p1 <- ggviolin(ps1.meta, x = \"group\", y = \"Shannon\",add = c(\"jitter\",\"boxplot\"), fill = \"group\", rotate = TRUE, palette = c(\"#a6cee3\", \"#b2df8a\", \"#fdbf6f\"))\n", "print(p1)\n", "p1 <- p1 + stat_compare_means(comparisons = bmi.pairs) \n", "print(p1)\n" ] }, { "cell_type": "code", "execution_count": null, "id": "impressive-stand", "metadata": {}, "outputs": [], "source": [ "############# Figure 1c. Time series profiles of Shannon diversity index\n", "### In R\n", "library(microbiome)\n", "library(ape)\n", "library(phyloseq)\n", "library(ggplot2)\n", "library(Rmisc)\n", "\n", "otus<-read.csv('OTUS.csv', header = TRUE, row.names = 1)\n", "otus=otus+1\n", "otus<-t(otus)\n", "taxmat<-read.csv(\"taxmat.csv\", header = TRUE, row.names = 1)\n", "samples<-read.csv(\"Sample_data.csv\", header = TRUE, row.names =1)\n", "taxmat<-as.matrix(taxmat)\n", "otus<-as.matrix(otus)\n", "\n", "OTU = otu_table(otus, taxa_are_rows = FALSE)\n", "TAX = tax_table(taxmat)\n", "SAM = sample_data(samples)\n", "physeq = phyloseq(OTU, TAX, SAM) ### generate phyloseq object\n", "random_tree = rtree(ntaxa(physeq), rooted=TRUE, tip.label=taxa_names(physeq))\n", "\n", "physeq = phyloseq(OTU, TAX, SAM, random_tree)\n", "\n", "df1<-samples[,2:3]\n", "df2 <- alpha(physeq, index = \"Shannon\")\n", "rownames(df1) <- c()\n", "rownames(df2) <- c()\n", "\n", "df<-cbind(df2,df1)\n", "tgc <- summarySE(df, measurevar=\"diversity_shannon\", groupvars=c(\"group\",\"time\"))\n", "pd <- position_dodge(0.1)\n", "\n", "d<-ggplot(tgc, aes(x=time, y=diversity_shannon, colour=group, group=group, fill=group)) + \n", " geom_errorbar(aes(ymin=diversity_shannon-se, ymax=diversity_shannon+se), colour=\"grey\", width=.3, position=pd) +\n", " geom_line(position=pd) +\n", " geom_point(position=pd, size=3, shape=21) + # 21 is filled circle\n", " xlab(\"Time (months)\") +\n", " ylab(\"Diversity\") +\n", " scale_colour_hue(name=\"Profiles\", # Legend label, use darker colors\n", " breaks=c(\"Stable\", \"Progressing\", \"Fluctuating\"),\n", " labels=c(\"Stable\", \"Progressing\", \"Fluctuating\"),\n", " l=50) + # Use darker colors, lightness=40\n", " expand_limits(y=2) + # Expand y range\n", " theme_bw() +\n", " theme(legend.justification=c(1,0),\n", " legend.position=c(1,0)) \n", " d" ] }, { "cell_type": "code", "execution_count": null, "id": "completed-rolling", "metadata": {}, "outputs": [], "source": [ "############# Figure 1d. Tracking rate of change from static timepoints. Using QIIME2.\n", "\n", "##### Convert OTUS table to biom format\n", "\n", "$ biom convert -i OTUS.txt -o OTUS_hdf5.biom --table-type=\"OTU table\" --to-hdf5\n", "\n", "##### In q2\n", "\n", "##### Import biom table\n", "\n", "qiime tools import \\\n", " --input-path OTUS_hdf5.biom \\\n", " --type 'FeatureTable[Frequency]' \\\n", " --input-format BIOMV100Format \\\n", " --output-path feature-table.qza\n", "\n", "##### ####### First differencing to track rate of change\n", "\n", "$ qiime diversity alpha --i-table feature-table.qza --p-metric shannon --o-alpha-diversity shannon.qza --verbose\n", "\n", "$ qiime longitudinal first-differences --m-metadata-file sample-metadata.tsv --m-metadata-file shannon.qza --p-state-column month --p-metric shannon_entropy --p-individual-id-column patient --p-replicate-handling random --o-first-differences shannon-first-differences.qza\n", "\n", "$ qiime longitudinal volatility --m-metadata-file sample-metadata.tsv --m-metadata-file shannon-first-differences.qza --p-default-group-column condition --p-state-column month --p-individual-id-column patient --o-visualization volatility-first-distances-shannon.qzv\n", "\n", "### View figure volatility-first-distances-shannon.qzv in QIIME2 View\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "id": "03a47c3e", "metadata": {}, "outputs": [], "source": [ "############# Figure 2. LEfSe analysis was run locally with default settings as follows: \n", "############# LEfSe_DNA_kraken.txt is \"merged_OTUs_table.tab\" with its headers modified to be read by LEfSe\n", "############# Headers have the following format:\n", "\n", "|Condition||stable||stable||...\n", "|Library||kraken||kraken||...\n", "|ID||16S-Sample-01||16S-Sample-02||...\n", "|d_Archaea||66||48||...\n", "\n", "$ python lefse-format_input.py LEfSe_DNA_kraken.txt LEfSe_DNA.in -c 1 -s 2 -u 3 -o 1000000\n", "$ python run_lefse.py LEfSe_DNA.in LEfSe_DNA.res -l 2.0\n", "$ python lefse-plot_res.py LEfSe_DNA.res LEfSe_DNA.svg \n" ] }, { "cell_type": "code", "execution_count": null, "id": "6218cf16", "metadata": {}, "outputs": [], "source": [ "############# Reconstructing ecological networks with the R package SpiecEasi ################\n", "\n", "##################################################\n", "# https://github.com/zdk123/SpiecEasi#installation\n", "##################################################\n", "\n", "\n", "### In R\n", "\n", "library(phyloseq)\n", "library(SpiecEasi)\n", "library(igraph)\n", "\n", "\n", "### Subset data for stable samples. Do the same for the other two conditions\n", "\n", "stable=subset_samples(physeq, group==\"stable\")\n", "\n", "\n", "### Reconstructing the ecological network\n", "\n", "se.mb.stable <- spiec.easi(stable, method='mb', lambda.min.ratio=1e-3,nlambda=10, pulsar.params=list(rep.num=20, ncores=35), verbose = TRUE) ### Lower lambda.min.ratio value if you get error messages\n", "\n", "ig2.mb <- adj2igraph(getRefit(se.mb.stable), vertex.attr=list(name=taxa_names(stable)))\n", "\n", "write_graph(ig2.mb, file='stable.gml', format=\"gml\") ############ Save network\n", "write_graph(ig2.mb, file='stable.csv', format=\"edgelist\") ############ Save edgelist\n", "\n", "plot_network(ig2.mb, stable, type='taxa') ### Plotting network\n", "\n", "\n", "#################### Figure 3a,b,c Network cartography #####################\n", "\n", "library(igraph)\n", "library(ggplot2)\n", "library(tidyverse)\n", "library(rnetcarto)\n", "\n", "####### Example fluctuating \n", "####### See https://cran.r-project.org/web/packages/tsna/vignettes/tsna_vignette.html where it describes the formats of PHStaticEdges and PHVertexAttributes\n", "\n", "PHStaticEdges <- read.csv('microbiome_StaticEdgelist_fluctuating.csv', stringsAsFactors = FALSE)\n", "PHVertexAttributes <- read.csv('microbiome_VertexAttributes_fluctuating.csv', stringsAsFactors = FALSE, header = TRUE)\n", "\n", "\n", "# Make and visualize our static network\n", "\n", "thenetwork <- network(\n", " PHStaticEdges,\n", " vertex.attr = PHVertexAttributes,\n", " vertex.attrnames = c(\"vertex.id\",\"name\"),\n", " directed = FALSE,\n", " bipartite = FALSE\n", ")\n", "\n", "g <- asIgraph(thenetwork)\n", "\n", "cartfluctuating<-netcarto(igraph::get.adjacency(g,sparse=FALSE))\n", "\n", "lapply(cartfluctuating, function(x) write.table( data.frame(x), 'Cartography_fluctuating.csv', append= T, sep=',' ))\n", "\n", "###### Plot results\n", "###### Delete last row of 'Cartography_fluctuating.csv'\n", "\n", "cartfl<-read.csv('Cartography_fluctuating.csv', header = TRUE)\n", "\n", "cartfl_2k <- cartfl %>%\n", " select(participation, connectivity, role, name)%>%\n", " drop_na()\n", "\n", "cartfl_2k %>%\n", " ggplot(aes(participation,connectivity,color=role)) +\n", " geom_text(aes(label = name), size = 3.5) +\n", " geom_point(alpha=0.5, size=2) +\n", " labs(y=\"Within-module connectivity z-score\", x=\"Participation coefficient (P)\")\n", "\n", "\n", "#################### Figure 3d,e,f Temporal network analysis with package tsna #####################\n", "\n", "############################################################################################################################\n", "# https://programminghistorian.org/en/lessons/temporal-network-analysis-with-r#packages-for-temporal-network-analysis\n", "# https://cran.r-project.org/web/packages/tsna/vignettes/tsna_vignette.html\n", "# https://rpubs.com/vitorks/723233\n", "############################################################################################################################\n", "\n", "library(sna)\n", "library(tsna)\n", "library(ndtv)\n", "\n", "######### Reconstruct the static network \n", "\n", "PHStaticEdges <- read.csv('microbiome_StaticEdgelist_progressing.csv', stringsAsFactors = FALSE)\n", "PHVertexAttributes <- read.csv('microbiome_VertexAttributes_progressing.csv', stringsAsFactors = FALSE, header = TRUE)\n", "\n", "thenetwork <- network(\n", " PHStaticEdges,\n", " vertex.attr = PHVertexAttributes,\n", " vertex.attrnames = c(\"vertex.id\",\"name\"),\n", " directed = FALSE,\n", " bipartite = FALSE\n", ")\n", "\n", "plot(thenetwork)\n", "\n", "### Import Temporal Network Data\n", "\n", "PHDynamicNodes <- read.csv('microbiome_DynamicNodes_progressing.csv', header = TRUE, stringsAsFactors = FALSE)\n", "PHDynamicEdges <- read.csv('microbiome_DynamicEdges_progressing.csv', header = TRUE, stringsAsFactors = FALSE)\n", "\n", "### Reconstruct the temporal network\n", "\n", "dynamicCollabs <- networkDynamic(\n", " thenetwork,\n", " edge.spells = PHDynamicEdges,\n", " vertex.spells = PHDynamicNodes\n", ")\n", "\n", "### Plot formation of edges over time\n", "\n", "plot(tEdgeFormation(dynamicCollabs, time.interval = 1))\n", "\n", "\n", "### Calculate and graph the rolling betweenness centralization of the network\n", "\n", "dynamicBetweenness <- tSnaStats(dynamicCollabs, snafun = \"centralization\", start = 0, end = 18, time.interval = 1, aggregate.dur = 2, FUN = \"betweenness\")\n", "\n", "plot(dynamicBetweenness)\n", "\n", "\n", "dynamicDegree <- tSnaStats( dynamicCollabs, snafun = \"centralization\", start = 0, end = 18, time.interval = 1, aggregate.dur = 2, FUN = \"degree\")\n", "\n", "plot(dynamicDegree)\n", "\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "id": "cardiovascular-appreciation", "metadata": {}, "outputs": [], "source": [ "#################### Dynamics of species over time. Figure 4a,b\n", "\n", "#######################################################################################################################################################\n", "# https://cran.r-project.org/web/packages/codyn/vignettes/Community_Stability_Metrics.html\n", "# https://cran.r-project.org/web/packages/codyn/vignettes/Temporal_Diversity_Indices.html\n", "#######################################################################################################################################################\n", "\n", "library(codyn)\n", "library(ggplot2)\n", "library(cowplot)\n", "library(ggpubr)\n", "\n", "########## Generate CODIN table from OTUS.csv in R\n", "\n", "OTUS1 <- read.csv(\"OTUS_codyn.csv\")\n", "\n", "codyn_table<-gather(OTUS1, key = \"Sample_ID\", value = \"measurement\", -Species)\n", "\n", "write.csv(codyn_table, file = \"OTUS_codyn.csv\")\n", "\n", "codyndata<-read.csv('OTUS_codyn.csv', header = TRUE, row.names = 1)\n", "\n", "### Species rank clocks (Figure 4a)\n", "\n", "### We selected the subset of species with >10% frequency in samples selected in Figure 1b.\n", "\n", "aggdat <- aggregate(abundance ~ species * time * replicate, \n", " data = subset(codyndata, \n", " species == \"Actinomyces_naeslundii\" |\n", " species == \"Actinomyces_oris\" |\n", " species == \"Actinomyces_sp._oral_taxon_170\" |\n", " species == \"Actinomyces_sp._oral_taxon_171\" |\n", " species == \"Actinomyces_sp._oral_taxon_448\" |\n", " species == \"Aggregatibacter_actinomycetemcomitans\"|\n", " species == \"Anaeroglobus_geminatus\" | \n", " species == \"Bacteroidetes_bacterium_oral_taxon_272\" |\n", " species == \"Bifidobacterium_longum\" | \n", " species == \"Brevundimonas_diminuta\" | \n", " species == \"Comamonas_testosteroni\"| \n", " species == \"Corynebacterium_durum\"| \n", " species == \"Dialister_invisus\" |\n", " species == \"Dialister_micraerophilus\" |\n", " species == \"Dietzia_sp._oral_taxon_368\" |\n", " species == \"Eggerthia_catenaformis\" |\n", " species == \"Fusobacteria_bacterium_oral_taxon_220\" |\n", " species == \"Fusobacterium_naviforme\" |\n", " species == \"Fusobacterium_necrophorum\" |\n", " species == \"Fusobacterium_nucleatum\" | \n", " species == \"Fusobacterium_periodonticum\"| \n", " species == \"Gemella_morbillorum\" |\n", " species == \"Haemophilus_haemolyticus\" |\n", " species == \"Haemophilus_sp._oral_taxon_036\" |\n", " species == \"Lactobacillus_casei\" |\n", " species == \"Lactobacillus_crispatus\" |\n", " species == \"Lactobacillus_fermentum\" |\n", " species == \"Lactobacillus_gasseri\" |\n", " species == \"Lactobacillus_panis\" |\n", " species == \"Lactobacillus_paracasei\" |\n", " species == \"Lactobacillus_rhamnosus\" |\n", " species == \"Lactococcus_lactis\" |\n", " species == \"Neisseria_elongata\" |\n", " species == \"Plectosphaerella_sp.\" |\n", " species == \"Polyporales_sp.\" |\n", " species == \"Prevotella_sp._oral_taxon_300\" |\n", " species == \"Prevotella_sp._oral_taxon_314\" |\n", " species == \"Schaalia_meyeri\" |\n", " species == \"Streptococcus_cristatus\" |\n", " species == \"Streptococcus_dysgalactiae\" |\n", " species == \"Streptococcus_mutans\" |\n", " species == \"Streptococcus_parasanguinis\" |\n", " species == \"Streptococcus_salivarius\" |\n", " species == \"Streptococcus_sinensis\" |\n", " species == \"Streptococcus_sp._oral_taxon_058\" |\n", " species == \"Streptococcus_sp._oral_taxon_064\" |\n", " species == \"Streptococcus_vestibularis\" |\n", " species == \"Tannerella_forsythia\" |\n", " species == \"Treponema_denticola\" |\n", " species == \"Veillonella_sp._oral_taxon_158\" |\n", " species == \"Xanthomonas_sp._oral_taxon_037\") ,\n", " FUN = mean)\n", "\n", "\n", "ggplot(aggdat, aes(time, abundance, color = species)) + \n", " geom_line(size = 1) + coord_polar() + theme_bw() + facet_wrap(~replicate) +\n", " ggtitle(\"Dominant species abundances\"\n", "\n", "\n", "### Mean rank shifts (Figure 4b)\n", "\n", "KNZ_rankshift <- rank_shift(df=codyndata, time.var = \"time\", species.var = \"species\", abundance.var = \"abundance\", replicate.var = \"replicate\")\n", "\n", "KNZ_rankshift$time <- abs(as.numeric((substr(KNZ_rankshift$year_pair, 3,9))))\n", "\n", "ggplot(KNZ_rankshift, aes(time, MRS, color=replicate)) + \n", " geom_line(size= 1) + theme_bw()\n", "\n", "### Rate of community change (Figure 4c and d)\n", "\n", "rate.res <- rate_change(codyndata, time.var= \"time\", species.var= \"species\", abundance.var= \"abundance\", replicate.var = \"replicate\")\n", "\n", "write.csv(rate.res, file = 'Rate_of_community_ change.csv')\n", "\n", "\n", "comm.res <- rate_change_interval(codyndata, time.var = \"time\",species.var = \"species\", abundance.var = \"abundance\",replicate.var = \"replicate\")\n", "\n", "ggplot(comm.res, aes(interval, distance, color = replicate)) + facet_wrap(~replicate) + \n", " geom_point() + theme_bw() + stat_smooth(method = \"lm\", se = F, size = 1)\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "id": "a57974e5", "metadata": {}, "outputs": [], "source": [ "############# Figure 5 a,b,c and Figure S5. Cluster different species' trajectories using the Dirichlet process Gaussian process mixture model (DPGP) software\n", "\n", "### First obtain the mean count for each species at each time point\n", "### Format of input is, where the first row represent the time points\n", "\n", " |0 |1 |2 |3 |4 |5 |6 |7\n", "species_0|1.49|0.36|2.44|0.15|2.14|4.58|1.60|0.73\n", "species_1|1.39|0.41|2.64|0.06|1.71|4.54|1.72|0.83\n", "species_2|1.63|0.56|2.92|0.25|2.13|4.22|1.95|0.88\n", "species_3|1.44|0.47|2.51|0.07|2.24|4.78|1.61|1.04\n", "….\n", "\n", "### In R\n", "\n", "Data<-read.table('condition_means.txt', header=TRUE, row.names=1) ### Condition: stable, progressing or fluctuating.\n", "\n", "### Standardized to zero mean and unit variance centering with 'scale()'.\n", "\n", "center_scale <- function(x) {\n", " scale(x, scale = TRUE)\n", "}\n", "centered<-center_scale(Data)\n", "write.table(centered, file = \"condition_centered.txt\", sep='\\t') ## This file is the input for the next step\n", "\n", "### In DPGP bin/ folder. DPGP can be obtained at:\n", "### https://github.com/PrincetonUniversity/DP_GP_cluster\n", "\n", "### Before proceding to the analsys you should remove all 0 rows\n", "\n", "$ bin/DP_GP_cluster.py -i condition_centered.txt -o ./stable --true_times --optimizer scg -a 0.5 -c MPEAR -n 1000 --save_cluster_GPs -p svg –plot\n", "\n" ] }, { "cell_type": "code", "execution_count": null, "id": "loose-cowboy", "metadata": {}, "outputs": [], "source": [ "############# Figure 5d,e,f. Measurements of stability, variance ratio, and synchrony. Community stability metrics\n", "### In R\n", "\n", "library(codyn)\n", "library(knitr)\n", "\n", "stability <- community_stability(codyndata, time.var = \"time\", abundance.var = \"abundance\", replicate.var = \"replicate\")\n", "write.csv(stability, file = 'Stability.csv')\n", "\n", "### Species covariance \n", "\n", "variance_ratio <- variance_ratio(df = codyndata, species.var = \"species\", time.var = \"time\", abundance.var = \"abundance\", bootnumber = 1000, replicate.var = \"replicate\")\n", "write.csv(variance_ratio, file = 'Variance_ratio.csv')\n", "\n", "variance_ratio_avgrep <- variance_ratio(codyndata, time.var = \"time\",species.var = \"species\", abundance.var = \"abundance\", bootnumber = 1000, replicate.var = \"replicate\", average.replicates = FALSE)\n", "write.csv(variance_ratio_avgrep, file = 'Variance_ratio_avgrep.csv')\n", "\n", "### Species synchrony\n", "\n", "synchrony_Loreau <- synchrony(df = codyndata, time.var = \"time\", species.var = \"species\", abundance.var = \"abundance\", replicate.var = \"replicate\")\n", "\n", "write.csv(synchrony_Loreau, file = 'Species_synchrony_Loreau.csv')\n" ] }, { "cell_type": "code", "execution_count": null, "id": "important-iraqi", "metadata": {}, "outputs": [], "source": [ "############# Figure 6. Phylodiversity accumulation and model fitting in the three datasets.\n", "### In R\n", "### Figure 5a,b,c\n", "\n", "library(\"ape\")\n", "library(\"ggplot2\")\n", "library(\"parallel\")\n", "source(\"assembly_model_functions.r\")\n", "\n", "tree_raw <- read.tree(\"ssu-otus-align.nwk\") ### Tree from the aligment of the sequences in the database\n", "otutable_raw <- read.table(\"OTUS.txt\", sep = '\\t', header = T, row.names = 1)\n", "metadata_raw <- read.table(\"sample-metadata.txt\", stringsAsFactors = F, sep='\\t', header = T, row.names = 1)\n", "\n", "\n", "inputs <- prep_check_inputs(\n", " otutable = otutable_raw,\n", " metadata = metadata_raw,\n", " md_order_col=4,\n", " tree = tree_raw,\n", " distmat=NULL, rarefy=TRUE, rare_depth=500,\n", " merge_tree_0dists=TRUE, drop_unobserved_otus=TRUE,\n", ")\n", "\n", "stable_inputs <- subset_inputs(\n", " inputs_list = inputs,\n", " sub_tf = inputs$metadata$group == \"stable\",\n", " name=\"stable\"\n", ")\n", "\n", "progressing_inputs <- subset_inputs(\n", " inputs_list = inputs,\n", " sub_tf = inputs$metadata$group == \"progressing\",\n", " name=\"progressing\"\n", ")\n", "\n", "fluctuating_inputs <- subset_inputs(\n", " inputs_list = inputs,\n", " sub_tf = inputs$metadata$group == \"fluctuating\",\n", " name=\"fluctuating\"\n", ")\n", "\n", "results_stable <- pd_assemb_workflow(\n", " inputs_i = stable_inputs,\n", " n_dispersions=1000,\n", " n_neutral_sim=1000,\n", " ncores = 40\n", ")\n", "\n", "results_progressing <- pd_assemb_workflow(\n", " inputs_i = progressing_inputs,\n", " n_dispersions=1000,\n", " n_neutral_sim=1000,\n", " ncores = 40\n", ")\n", "\n", "results_fluctuating <- pd_assemb_workflow(\n", " inputs_i = fluctuating_inputs,\n", " n_dispersions=1000,\n", " n_neutral_sim=1000,\n", " ncores = 40\n", ")\n", "\n", "plot_dispersion_ests_violin(list(results_stable, results_progressing,results_fluctuating))\n", "\n", "### Figure 6d. Testing against alternative “individual null”. Example ‘progressing’ \n", "### In R\n", "\n", "source(\"alt_null_functions.r\") ## This r function is included in the submission.\n", "Nperm <- 5000\n", "\n", "# define m, which is midpoint sample index. See paper for more details.\n", "m <- round(ncol(progressing_inputs$otutable)/2)\n", "\n", "# get vector of nperm null PDs\n", "ind_nuls <- replicate(nperm, indiv_null(progressing_inputs$otutable, progressing_inputs$tree, nsamp=m))\n", "\n", "# get empirical PD\n", "emp <- one_sample_pd(progressing_inputs$cum_otutable[,m], ids=rownames(progressing_inputs$cum_otutable), tree=progressing_inputs$tree)\n", "\n", "# calculate P-value\n", "pval <- (sum(ind_nuls < emp) + 1) / length(ind_nuls)\n", "pval" ] }, { "cell_type": "code", "execution_count": null, "id": "adopted-eight", "metadata": {}, "outputs": [], "source": [ "############# Figure S2 ARIMA\n", "### Augmented Dickey-Fuller Test of stationarity \n", "### In R\n", "\n", "library(tseries)\n", "library(astsa)\n", "library(aTSA)\n", "\n", "x<-read.csv('stable.csv', header = TRUE) ### stable.csv contains the observed values of CAL\n", "y<-x[,4] ### select columns with the average values\n", "y<-y[!is.na(y)] ### remove NAs\n", "adf.test(y, alternative = 'stationary') ## p > 0.05 is non-stationary\n", "\n", "\n", "### We transformed the series into stationary on the values for non-stationary series by subtracting CALt-1 from CALt for all values t. This technique is called differencing and can be done with the ‘diff’ function of the aTSA package.\n", "\n", "\n", "### Stationary time-series. We applied the ARIMA model for stationary series directly. \n", "\n", "x<-read.csv('boxplot_stable.csv', header = TRUE)\n", "y<-x[,4] ### select columns with the average values\n", "y<-y[!is.na(y)]\n", "y<-ts(y)\n", "plot(y)\n", "gtemp=y\n", "sarima.for(gtemp,5,1,1,1) ## ARIMA model is (1,1,1)\n", "\n", "### Non-stationary time-series\n", "\n", "x<-read.csv('boxplot_progressing.csv', header = TRUE)\n", "y<-x[,4] ### select columns with the average values\n", "y<-y[!is.na(y)]\n", "y=diff(y,differences = 1) ## We transformed the series into stationary on the values for non-stationary series by subtracting CALt-1 from CALt for all values t. This technique is called differencing\n", "y<-ts(y)\n", "plot(y)\n", "gtemp=y\n", "sarima.for(gtemp,5,0,1,1)\n" ] }, { "cell_type": "code", "execution_count": null, "id": "facial-contents", "metadata": {}, "outputs": [], "source": [ "############# Figure S3a. Alpha diversity \n", "### In R\n", "\n", "library(microbiomeSeq)\n", "library(phyloseq)\n", "\n", "otus<-read.csv('OTUS.csv', header = TRUE, row.names = 1)\n", "otus = t(otus)\n", "taxmat<-read.csv(\"taxmat.csv\", header = TRUE, row.names = 1)\n", "samples<-read.csv(\"pitlatrine2.csv\", header = TRUE, row.names =1)\n", "\n", "taxmat<-as.matrix(taxmat)\n", "otus<-as.matrix(otus)\n", "\n", "OTU = otu_table(otus, taxa_are_rows = FALSE)\n", "TAX = tax_table(taxmat)\n", "SAM = sample_data(samples)\n", "\n", "physeq = phyloseq(OTU, TAX, SAM) ### generate phyloseq object\n", "\n", "random_tree = rtree(ntaxa(physeq), rooted=TRUE, tip.label=taxa_names(physeq))\n", "\n", "physeq = phyloseq(OTU, TAX, SAM, random_tree)\n", "\n", "\n", "sample_data(physeq)$group = as.character(sample_data(physeq)$group)\n", "get_variable(physeq,\"group\")\n", "\n", "p <- plot_anova_diversity(physeq, method = c(\"richness\", \"fisher\"), grouping_column = \"condition\", pValueCutoff = 0.05)\n", "print(p)\n", "\n", "############# Figure S3b. Alpha diversity \n", "p <- plot_anova_diversity(physeq, method = c(\"shannon\"), grouping_column = \"grouptime\", pValueCutoff = 0.05)\n", "print(p)\n" ] }, { "cell_type": "code", "execution_count": null, "id": "b3688a48", "metadata": {}, "outputs": [], "source": [ "############# Figure S4. Ordination and β-dispersion. \n", "### In R\n", "\n", "library(microbiome) # data analysis and visualisation\n", "library(phyloseq) # also the basis of data object. Data analysis and visualisation\n", "library(RColorBrewer) # nice color options\n", "library(ggpubr) # publication quality figures, based on ggplot2\n", "library(dplyr) # data handling\n", "library(ape)\n", "library(vegan)\n", "\n", "otus<-read.csv('OTUS.csv', header = TRUE, row.names = 1)\n", "otus = t(otus)\n", "taxmat<-read.csv(\"taxmat.csv\", header = TRUE, row.names = 1)\n", "samples<-read.csv(\"Sample_data.csv\", header = TRUE, row.names =1)\n", "samples$time=factor(samples$time)\n", "\n", "taxmat<-as.matrix(taxmat)\n", "otus<-as.matrix(otus)\n", "\n", "\n", "OTU = otu_table(otus, taxa_are_rows = FALSE)\n", "TAX = tax_table(taxmat)\n", "SAM = sample_data(samples)\n", "\n", "physeq = phyloseq(OTU, TAX, SAM) ### generate phyloseq object\n", "\n", "random_tree = rtree(ntaxa(physeq), rooted=TRUE, tip.label=taxa_names(physeq))\n", "\n", "physeq = phyloseq(OTU, TAX, SAM, random_tree)\n", "\n", "\n", "ps1=physeq\n", "ps1.rel <- microbiome::transform(ps1, \"compositional\") ### Normalized counts\n", " \n", "######################## Ordination\n", "\n", "### For Bray-Curtis, Jaccard substitute the method\n", "\n", "ordu.wt.uni <- ordinate(ps1.rel , \"PCoA\", \"unifrac\", weighted=T)\n", "\n", "wt.unifrac <- plot_ordination(ps1.rel, \n", " ordu.wt.uni, color=\"group\") \n", "wt.unifrac <- wt.unifrac + ggtitle(\"Weighted UniFrac\") + geom_point(size = 3)\n", "wt.unifrac <- wt.unifrac + theme_classic() + scale_color_brewer(\"Location\", palette = \"Set2\")\n", "print(wt.unifrac)\n", "\n", "print(wt.unifrac + stat_ellipse())\n", "\n", "\n", "############################## Permanova Unifrac ##############################\n", "\n", "metadf <-as.data.frame(samples)\n", "unifrac.dist <- UniFrac(ps1.rel, \n", " weighted = TRUE, \n", " normalized = TRUE, \n", " parallel = FALSE, \n", " fast = TRUE)\n", "\n", "\n", "permanova <- adonis2(unifrac.dist ~ group, data = metadf)\n", "\n", "permanova\n", "\n", "############## Permanova in the case of Bray-Curtis and Jaccard ################\n", "\n", "x=otu_table(ps1.rel)\n", "\n", "bray.dist = vegdist(x, method=\"bray\", binary=FALSE)\n", "\n", "permanova <- adonis2(bray.dist ~ group, data = metadf)\n", "\n", "permanova\n", "\n", "#####\n", "\n", "x=otu_table(ps1.rel)\n", "\n", "jaccard.dist = vegdist(x, method=\"jaccard\", binary=FALSE)\n", "\n", "permanova <- adonis2(jaccard.dist ~ group, data = metadf)\n", "\n", "permanova\n", "\n", "\n", "\n", "############################## Beta-dispersion ###############################################\n", "\n", "ps.disper <- betadisper(bray.dist, metadf$group,type = \"centroid\", bias.adjust = TRUE)\n", "permutest(ps.disper, pairwise = TRUE)\n", "\n", "ps.disper <- betadisper(jaccard.dist, metadf$group,type = \"centroid\", bias.adjust = TRUE)\n", "permutest(ps.disper, pairwise = TRUE)\n", "\n", "ps.disper <- betadisper(unifrac.dist, metadf$group, type = \"centroid\", bias.adjust = TRUE) \n", "permutest(ps.disper, pairwise = TRUE)\n", "\n", "\n", "\n", "\n" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.13" } }, "nbformat": 4, "nbformat_minor": 5 }