“Publication Figure 4”

百度云盘链接: https://pan.baidu.com/s/15g7caZp354zIWktpnWzWhQ

提取码: 4sh7

Libraries

Standard Import

library(tidyverse)
library(cowplot)
library(scales)
library(ggpubr)

Special

library(caret)
library(plotROC)
library(tidymodels)
library(grid)
library(sjPlot)

Paths

bin_dir = '../bin/'
data_dir = '../data/'
results_dir = '../results/'

Custom Scripts / Theme

source(paste0(bin_dir, 'theme_settings.R'))
source (paste0(bin_dir, 'utilities.R'))
source(paste0(bin_dir, 'species_competition_functions.R'))
source(paste0(bin_dir, 'distmat_utils.R'))
source(paste0(bin_dir, 'analysis_metadata.R'))
date <- format(Sys.time(), "%d%m%y")

Import Tables

Metadata

samples.metadata <- read_tsv(paste0(data_dir, 'samples.metadata.tsv'))
microbe.taxonomy <- read_tsv(paste0(data_dir, 'microbe.taxonomy.tsv'))
microbe.metadata <- microbe.taxonomy %>% right_join(., read_tsv(paste0(data_dir, 'microbe.metadata.tsv'))) 

Taxonomy

MetaPhlAn2 tables
combine with taxonomy

mp_species.long <- microbe.taxonomy %>% right_join(., read_tsv(paste0(data_dir, 'samples.mp_profiles.species.tsv')), by = 'species') %>% left_join(., samples.metadata, by = 'Name')

rCDI subset
annotate with metadata

mp_species.rcdi.long <- mp_species.long %>%filter(Study %in% c(rcdi.studies)) 

SameStr Data

SameStr Case-Summary table

samples.sstr_data <- read_tsv(paste0(data_dir, 'samples.sstr_data_wmetadata.tsv')) %>% left_join(., microbe.taxonomy, by = 'species')

Case Summary

SameStr Case-Summary table

sstr_cases <- read_tsv(paste0(data_dir, 'samples.case_summary.tsv')) %>% left_join(., microbe.taxonomy, by = 'species')

Functions

precision <- function(matrix) {tp <- matrix[2, 2]fp <- matrix[1, 2]return (tp / (tp + fp))
}
recall <- function(matrix) {tp <- matrix[2, 2]fn <- matrix[2, 1]return (tp / (tp + fn))
}

Set Figure

fig = paste0('Fig_4.')

Identify Individuals

Identifying related sample pairs coming from the same individual (controls) or from related biological context (rCDI) based on the number and fraction of shared family, genera, species and strains.

Format

Generate MetaPhlAn co-occurrence table, taxa counts

cooccurrence <- mp_cooccurrence(mp_species.long)
taxa_counts <- cooccurrence$mp_taxa_counts
taxa_cooccurrence <- cooccurrence$mp_taxa_table.long

Merge with SameStr strain-level data

sstr <-samples.sstr_data %>% group_by(Study.pair, Name.row, Name.col, Days_Since_FMT.row, Days_Since_FMT.col,Case_Name.pair, Case_Name.row, Case_Name.col, Same_Case.label, Related.label, Related.bool, Pair_Type) %>%summarize(.groups = 'drop',n.analyzed_strains = sum(overlap > min_overlap, na.rm = T),n.shared_strains.mvs = sum(distance.mvs > min_similarity & overlap > min_overlap, na.rm = T),n.shared_strains.cvs = sum(distance.cvs > min_similarity & overlap > min_overlap, na.rm = T)) %>%# tag successful fmt casesmutate(fmt_success = !Case_Name.pair %in% cases_failed) %>% mutate(Study.row = str_split_fixed(Name.row, pattern = '_', 2)[,1], Study.col = str_split_fixed(Name.col, pattern = '_', 2)[,1]) %>% mutate(Related.text = ifelse(Related.bool, 'Related', 'Unrelated')) %>% rowwise() %>%mutate(Sample_Pair = paste0(sort(c(Name.row, Name.col)), collapse = ',')) %>% ungroup() %>% left_join(., taxa_cooccurrence, by = 'Sample_Pair') %>% left_join(., taxa_counts, by = c('Name.row' = 'Sample')) %>% left_join(., taxa_counts, by = c('Name.row' = 'Sample'), suffix = c('.row','.col')) %>% mutate(n.not_shared_species = (n.species.row - n.shared_species) + (n.species.col - n.shared_species), n.not_shared_genus = (n.genus.row - n.shared_genus) + (n.genus.col - n.shared_genus), n.not_shared_family = (n.family.row - n.shared_family) + (n.family.col - n.shared_family), n.total_species = n.shared_species + n.not_shared_species, n.total_genus = n.shared_genus + n.not_shared_genus, n.total_family = n.shared_family + n.not_shared_family, f.shared_strains.mvs = replace_na(n.shared_strains.mvs / n.analyzed_strains, 0), f.shared_strains.cvs = replace_na(n.shared_strains.cvs / n.analyzed_strains, 0), f.shared_species = n.shared_species / n.total_species,f.shared_genus = n.shared_genus / n.total_genus, f.shared_family = n.shared_family / n.total_family)

Subset rCDI and Controls

# filter only healthy controls
sstr.control <-sstr %>% filter(Study.row %in% control.studies &Study.col %in% control.studies)sstr.rcdi <- sstr %>% filter(Study.row %in% rcdi.studies &Study.col %in% rcdi.studies)sstr.louiss <-sstr %>% filter(Study.row == 'LOUISS',Study.col == 'LOUISS')

Logistic Regression Model

Train on Control

Model on train data of healthy controls

set.seed(50)
# 60% train / 40% test
trainIndex <- caret::createDataPartition(sstr.control$Related.bool, p = .6, list = FALSE, times = 1)[,1]
train <- sstr.control[ trainIndex,]
test <- sstr.control[-trainIndex,]fit.norm.strain.all <-glm(Related.bool ~ n.total_species + f.shared_strains.mvs + n.shared_strains.mvs +f.shared_species + n.shared_species +f.shared_genus + n.shared_genus +f.shared_family + n.shared_family, data = train, family = binomial(link = 'logit'))# normal models
fit.norm.strain <-glm(Related.bool ~ f.shared_strains.mvs + n.shared_strains.mvs + n.analyzed_strains, data = train, family = binomial(link = 'logit'))
fit.norm.species <-glm(Related.bool ~ f.shared_species + n.shared_species + n.total_species, data = train, family = binomial(link = 'logit'))
fit.norm.genus <-glm(Related.bool ~ f.shared_genus + n.shared_genus + n.total_genus, data = train, family = binomial(link = 'logit'))
fit.norm.family <-glm(Related.bool ~ f.shared_family + n.shared_family + n.total_family, data = train, family = binomial(link = 'logit'))

Test on Control

prediction on test data of healthy controls

test$pred.norm.strain <- predict(fit.norm.strain, test, type = 'response')
test$pred.norm.strain.all <- predict(fit.norm.strain.all, test, type = 'response')
test$pred.norm.species <- predict(fit.norm.species, test, type = 'response')
test$pred.norm.genus <- predict(fit.norm.genus, test, type = 'response')
test$pred.norm.family <- predict(fit.norm.family, test, type = 'response')test.long <-test %>% pivot_longer(names_to = 'model', values_to = 'M', cols = c('pred.norm.strain',# pred.norm.strain.all,'pred.norm.species','pred.norm.genus','pred.norm.family'))

Performance

PR with yardstick from tidymodels

options(yardstick.event_first = FALSE)c.d1 <-test.long %>% mutate(model = gsub(model, pattern = 'pred.norm.', replacement = ''),model = fct_relevel(model, 'family', 'genus', 'species', 'strain'))c.d2 <-c.d1 %>% dplyr::group_by(model) %>% dplyr::mutate(Related.bool = factor(Related.bool)) %>%yardstick::pr_curve(data = ., truth = Related.bool, M)c.d3 <-c.d1 %>% dplyr::group_by(model) %>% dplyr::mutate(Related.bool = factor(Related.bool)) %>%  pr_auc(data = ., truth = Related.bool, M, na_rm = T)

ROC with plotroc

x1 = .95
x2 = .77
yy = .17
p.expand <- rep(0.01, 2)# auROC
control.auROC <-c.d1 %>% ggplot(aes(m = M, d = as.numeric(Related.bool), color = model)) + geom_roc(n.cuts = 0, labels = F) +theme_cowplot() +coord_fixed() + scale_color_manual(values = colors.discrete) + scale_x_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand) + scale_y_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand, sec.axis = dup_axis()) + geom_abline(slope = 1, linetype = 'dashed') + labs(x = 'False Positive', y = 'True Positive') +theme(legend.position = 'none',axis.line.y.right = element_blank(),axis.ticks.y.right = element_blank(),axis.text.y.right = element_blank(),axis.title.y.left = element_blank()# axis.text.x = element_blank(), # axis.ticks.x = element_blank(),# axis.line.x = element_blank())control.auROC.values <- calc_auc(control.auROC)$AUC
control.auPR.values <- c.d3$.estimatecontrol.auROC.table <-control.auROC + theme(legend.justification = c(0, 1), legend.position = c(.35, .38), legend.title = element_blank()) +annotate(geom = 'text', label = 'bold("auPR")', x = x1, y = .375, parse = T) +annotate(geom = 'text', label = 'bold("auROC")', x = x2, y = .375, parse = T) +annotate(geom = 'text', label = paste0(round(control.auROC.values, 3), collapse = '\n'),x = x2, y = yy, parse = F) +annotate(geom = 'text', label = paste0(round(control.auPR.values, 3), collapse = '\n'),x = x1, y = yy, parse = F)# auPR (precision-recall plot)
control.auPR <-c.d2 %>% ggplot(aes(y = precision, x = recall, col = model)) + geom_line(size = 1) + coord_fixed() + theme_cowplot() +scale_color_manual(values = colors.discrete) + scale_x_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand) + scale_y_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand, sec.axis = dup_axis()) + labs(x = 'Recall', y = 'Precision') + geom_abline(intercept = 1, slope = -1, linetype = 'dashed') + theme(legend.position = 'none',axis.line.y.right = element_blank(),axis.ticks.y.right = element_blank(),axis.text.y.right = element_blank(),axis.title.y.left = element_blank())control.auROC
control.auPR

Confusion Matrix

Confusion Matrix using a conservative threshold, showing that the simple logistic regression classifier achieves ~90% accuracy for within-cases comparisons, almost 100% accuracy for distinct subjects (total FP = 2), and has most difficulty with assigning the the correct label to related pre/post-FMT samples and distinct individuals that were treated with stool from the same donor.

# inspect false calls
sstr.control.confusion <-test %>% mutate(Predicted = pred.norm.strain > 8.713023e-08,TP = Predicted & Related.bool,FP = Predicted & !Related.bool,FN = !Predicted & Related.bool,TN = !Predicted & !Related.bool) %>% group_by(Pair_Type, Related.label, fmt_success) %>% summarize(TP = sum(TP, na.rm = T),FP = sum(FP, na.rm = T),FN = sum(FN, na.rm = T),TN = sum(TN, na.rm = T),.groups = 'drop') %>% mutate(Acc = (TP + TN) / (TP + TN + FP + FN)) sstr.control.confusion

Test on rCDI

Predict using control-trained model on FMT sample pairs, including donor/post-FMT, recipient/post-FMT.

# prediction
sstr.rcdi$pred.norm.strain <- predict(fit.norm.strain, sstr.rcdi, type = 'response')
sstr.rcdi$pred.norm.strain.all <- predict(fit.norm.strain.all, sstr.rcdi, type = 'response')
sstr.rcdi$pred.norm.species <- predict(fit.norm.species, sstr.rcdi, type = 'response')
sstr.rcdi$pred.norm.genus <- predict(fit.norm.genus, sstr.rcdi, type = 'response')
sstr.rcdi$pred.norm.family <- predict(fit.norm.family, sstr.rcdi, type = 'response')sstr.rcdi.long <-sstr.rcdi %>% pivot_longer(names_to = 'model', values_to = 'M', cols = c('pred.norm.strain',# pred.norm.strain.all,'pred.norm.species','pred.norm.genus','pred.norm.family'))

Performance

PR curve

options(yardstick.event_first = FALSE)r.d1 <-sstr.rcdi.long %>% mutate(model = gsub(model, pattern = 'pred.norm.', replacement = ''),model = fct_relevel(model, 'family', 'genus', 'species', 'strain'))r.d2 <-r.d1 %>% dplyr::ungroup() %>% dplyr::group_by(model) %>%dplyr::mutate(Related.bool = factor(Related.bool)) %>%pr_curve(data = ., truth = Related.bool, M)r.d3 <-r.d1 %>% dplyr::group_by(model) %>% dplyr::mutate(Related.bool = factor(Related.bool)) %>%  pr_auc(data = ., truth = Related.bool, M, na_rm = T)

ROC

x1 = .95
x2 = .77
yy = .17
p.expand <- rep(0.01, 2)# auROC
rcdi.auROC <-r.d1 %>% ggplot(aes(m = M, d = as.numeric(Related.bool), color = model)) + geom_roc(n.cuts = 0, labels = F) +theme_cowplot() +coord_fixed() + scale_color_manual(values = colors.discrete) + scale_x_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand) +scale_y_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand, sec.axis = dup_axis()) +geom_abline(slope = 1, linetype = 'dashed') + labs(x = 'False Positive', y = 'True Positive') + theme(aspect.ratio = 1,legend.position = 'none',axis.line.y.right = element_blank(),axis.ticks.y.right = element_blank(),axis.text.y.right = element_blank(),axis.title.y.left = element_blank())rcdi.auROC.values <- calc_auc(rcdi.auROC)$AUC
rcdi.auPR.values <- r.d3$.estimate# auPR (precision-recall plot)
rcdi.auPR <-r.d2 %>% ggplot(aes(y = precision, x = recall, col = model)) + geom_line(size = 1) + coord_fixed() + theme_cowplot() +scale_color_manual(values = colors.discrete) + scale_x_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand) +scale_y_continuous(labels = percent_format(), limits = c(0, 1), expand = p.expand, sec.axis = dup_axis()) +labs(x = 'Recall', y = 'Precision') + geom_abline(intercept = 1, slope = -1, linetype = 'dashed') + theme(aspect.ratio = 1,legend.position = 'none',axis.line.y.right = element_blank(),axis.ticks.y.right = element_blank(),axis.text.y.right = element_blank(),axis.title.y.left = element_blank())rcdi.auROC
rcdi.auPR

Confusion Matrix

Confusion Matrix using a conservative threshold, showing that the simple logistic regression classifier achieves ~90% accuracy for within-cases comparisons, almost 100% accuracy for distinct subjects (total FP = 2), and has most difficulty with assigning the the correct label to related pre/post-FMT samples and distinct individuals that were treated with stool from the same donor.

# inspect false calls
sstr.rcdi.confusion <-sstr.rcdi %>% mutate(Predicted = pred.norm.strain > 0.001774427,TP = Predicted & Related.bool,FP = Predicted & !Related.bool,FN = !Predicted & Related.bool,TN = !Predicted & !Related.bool) %>% group_by(Pair_Type, Related.label, fmt_success) %>% summarize(TP = sum(TP, na.rm = T),FP = sum(FP, na.rm = T),FN = sum(FN, na.rm = T),TN = sum(TN, na.rm = T),.groups = 'drop') %>% mutate(Acc = (TP + TN) / (TP + TN + FP + FN)) sstr.rcdi.confusion

Write to File

sstr.rcdi.confusion %>%  write_tsv(., paste0(results_dir, fig, 'ConfusionMatrix.rCDI.tsv'))

Scatter Plot

r.d1 %>% ggplot(aes(n.shared_strains.mvs, n.shared_species)) + geom_point(aes(fill = Related.label), size = 2, shape = 21) + scale_fill_manual(values = colors.discrete) + theme_cowplot() + theme(legend.title = element_blank()) +labs(x = 'Shared Strains [n]',y = 'Shared Species [n]') r.d1 %>% ggplot(aes(f.shared_strains.mvs, f.shared_species)) + geom_point(aes(fill = Related.label), size = 2, shape = 21) + geom_abline(slope = 1, intercept = 0, linetype = 'dashed') + scale_x_continuous(labels = percent_format(), limits = c(0, 1)) +scale_y_continuous(labels = percent_format(), limits = c(0, 1)) +theme_cowplot() + theme(aspect.ratio = 1, legend.title = element_blank()) + scale_fill_manual(values = colors.discrete) + labs(x = 'Shared Strains [%]',y = 'Shared Species [%]')

Model Summary

Exporting Table

model_comparison <- tibble(model = c('family', 'genus', 'species', 'strain'),control.auROC = control.auROC.values, control.auPR = control.auPR.values, rcdi.auROC = rcdi.auROC.values, rcdi.auPR = rcdi.auPR.values, )output_name = 'RelatedSubject.ModelPerformance'
sjPlot::tab_df(model_comparison, digits = 4,alternate.rows = F,file = paste0(results_dir, fig, output_name, '.doc'), title = 'Logistic Regression Model Performance')

Exporting Plots

ww = 3.25
hh = 3
# legend
legend = cowplot::get_legend(rcdi.auROC + theme(legend.position = 'left',legend.title = element_blank()))
output_name = 'RelatedSubject'
ggsave(legend, device = 'pdf', filename = paste0(results_dir, fig, output_name, '.Legend.pdf'), dpi = 300, width = 1, height = .9)# control
output_name = 'RelatedSubject.Control'
ggsave(control.auROC, device = 'pdf', filename = paste0(results_dir, fig, output_name, '.auROC.pdf'), dpi = 300, width = ww, height = hh)ggsave(control.auPR, device = 'pdf', filename = paste0(results_dir, fig, output_name, '.auPR.pdf'), dpi = 300, width = ww, height = hh)# rcdi
output_name = 'RelatedSubject.rCDI'
ggsave(rcdi.auROC, device = 'pdf', filename = paste0(results_dir, fig, output_name, '.auROC.pdf'), dpi = 300, width = ww, height = hh)ggsave(rcdi.auPR, device = 'pdf', filename = paste0(results_dir, fig, output_name, '.auPR.pdf'), dpi = 300, width = ww, height = hh)

Strain-level Source (Control)

Strain-level plot of per-Case second time point rel. abund. by source

Format

sstr_cases.control.long <- sstr_cases %>%filter(Study_Type == 'Control') %>% tag_species_competition(.) %>% mutate(n = 1) %>% mutate(source = ifelse(grepl(species, pattern = 'unclassified'),  'Unclassified', source)) %>% mutate(source = case_when(analysis_level == 'species' & source == 'self' ~ 'Initial Sample Sp.',analysis_level == 'species' & source == 'unique' ~ 'New Sp.',T ~ source)) %>% group_by(Study_Type, Case_Name, source, Days_Since_FMT.post, fmt_success.label) %>% summarize(n = sum(n, na.rm = T),rel_abund = sum(rel_abund.post, na.rm = T) / 100,.groups = 'drop') %>% group_by(Case_Name, Days_Since_FMT.post) %>% mutate(f = n / sum(n, na.rm = T)) %>%ungroup() %>% dplyr::select(-n) %>% group_by(Case_Name) %>% mutate(source = case_when(source == 'Unclassified' ~ 'Unclassified Sp.',source == 'same_species' ~ 'Same Sp.', source == 'unique' ~ 'New',source == 'self' ~ 'Initial Sample',T ~ source)) %>% pivot_wider(names_from = 'source', values_from = c('rel_abund','f'), names_sep = '___') %>% mutate_at(.vars = vars(contains('___')),.funs = funs(replace_na(., 0))) %>%pivot_longer(cols = contains("___"), names_to = c("metric", "source"),names_sep = '___', values_drop_na = F)

Statistics
Retained (strain-level) from previous time point: 73.1 ± 18.3%

sstr_cases.control.long  %>% filter(metric == 'rel_abund') %>% filter(Days_Since_FMT.post <= 373) %>% # ~ same time range as last rcdigroup_by(Study_Type, Case_Name) %>% filter(dense_rank(Days_Since_FMT.post) == max(dense_rank(Days_Since_FMT.post))) %>% ungroup() %>% group_by(Study_Type, source) %>% summarize(rel_abund.mean = mean(value, na.rm = T), rel_abund.sd = sd(value, na.rm = T),.groups = 'drop') sstr_cases.control.long  %>% filter(metric == 'f') %>% filter(Days_Since_FMT.post <= 373) %>% # ~ same time range as last rcdigroup_by(Study_Type, Case_Name) %>% filter(dense_rank(Days_Since_FMT.post) == max(dense_rank(Days_Since_FMT.post))) %>% ungroup() %>% group_by(Study_Type, source) %>% summarize(fraction.mean = mean(value, na.rm = T), fraction.sd = sd(value, na.rm = T),.groups = 'drop')

Bar Chart, per Case

Control - Bar charts showing second time point rel. abund. of sets of co-occurring strains in first and second time points of controls.

strain_order <- c('Initial Sample', 'Initial Sample Sp.', 'New', 'New Sp.', 'Same Sp.')
strain_colors <- c(colors.discrete[c(8,3, 7,2, 10,5)])
use_vars = list(strain_order, strain_colors)plot.control <-sstr_cases.control.long  %>% filter(metric == 'rel_abund') %>% filter(Days_Since_FMT.post <= 373) %>% # ~ same time range as last rcdiungroup() %>% mutate(ordering = as.numeric(as.factor(Case_Name)) + (Days_Since_FMT.post/1000)) %>% ggplot(aes(fct_reorder(paste0('D', Days_Since_FMT.post, ' | ', str_remove(Case_Name, pattern = 'Case_')), ordering),y = value,fill = fct_relevel(source, use_vars[[1]]))) + geom_bar(stat = 'identity', position = position_fill(), width = 1, col = 'black') +theme_cowplot() + scale_y_continuous(labels = percent_format(),breaks = c(1, .75, .5, .25, 0),expand = c(0,0)) +scale_fill_manual(values = strain_colors, guide = guide_legend(reverse = TRUE)) + labs(y = 'Rel. Abund.') + theme(axis.title.x = element_blank(),axis.line.y = element_blank(),axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5, size = 7.5),legend.title=element_blank(),legend.position = 'top') 
plot.control

Exporting plots

output_name = 'PostSource.Cases.Control.Strain'
ggsave(plot.control + theme(legend.position = 'none'), device = 'pdf', dpi = 300, width = 25, height = 3.5,filename = paste0(results_dir, fig, output_name, '.Plot.pdf'))

Area Chart, Across Cases

Control - Area chart showing second time point rel. abund. of sets of co-occurring strains in first and second time points of controls based on glm 2nd binomial model across control cases

strain_order <- c('Initial Sample', 'Initial Sample Sp.', 'New', 'New Sp.', 'Same Sp.')
strain_colors <- c(colors.discrete[c(8,3, 7,2, 10,5)])
use_vars = list(strain_order, strain_colors)pseudo = 1e-6
plot <-sstr_cases.control.long  %>% filter(Days_Since_FMT.post <= 373) %>% # ~ same time range as last rcdiggplot(aes(x = Days_Since_FMT.post, y = value, fill = fct_relevel(source, use_vars[[1]]))) +  stat_smooth(geom = 'area',method = 'glm', colour = 'black', size = 0.25,position = 'fill',method.args=list(family='binomial')) +facet_grid(scales = 'free_x', space = 'free_x',cols = vars(''),rows = vars(ifelse(grepl(metric, pattern = '^f'), 'Spp. Fraction', 'Rel. Abund.'))) +theme_cowplot() + scale_y_continuous(labels = percent_format(),breaks = c(1, .75, .5, .25, 0),expand = c(0,0), ) +scale_x_continuous(trans = pseudo_log_trans(0.5, 10), breaks = c(pseudo, 7, 14, 35, 84, 365),expand = c(0,0)) + scale_fill_manual(values = use_vars[[2]], guide = guide_legend(reverse = TRUE)) + # labs(x = 'Days Since\nInitial Sampling') + labs(x = 'Days between\nSample Collection') + theme(axis.title.y = element_blank(), plot.margin = unit(c(0, 0, 0, 0), "cm"), strip.background = element_blank(), strip.text = element_text(size = 14), panel.spacing.y = unit(.5, "cm"),panel.spacing.x = unit(.35, "cm"),legend.title=element_blank(),legend.margin=ggplot2::margin(l = 0.1, unit='cm'))plot + theme(legend.position = 'none')
legend <- cowplot::get_legend(plot)grid.newpage()
grid.draw(legend)

Histogram of available samples for modelling

plot.histogram <-sstr_cases.control.long %>% filter(metric == 'f') %>% group_by(Study_Type, Case_Name, Days_Since_FMT.post) %>% summarize(value = sum(value, na.rm = T)) %>% ggplot(aes(Days_Since_FMT.post)) + geom_histogram(fill = 'black', bins = 10) +facet_grid(scales = 'free_x', space = 'free_x', cols = vars(fct_relevel(Study_Type, 'rCDI', 'Control')),rows = vars('n')) +scale_x_continuous(trans = pseudo_log_trans(0.5, 10), breaks = c(pseudo, 7, 14, 30, 90, 365),expand = c(0,0)) + scale_y_continuous(breaks = c(0, 12, 25), expand = c(0,0),) +theme_cowplot() + theme(strip.background = element_blank(), strip.text.x = element_blank(), axis.title = element_blank(), axis.line.x = element_line(size = 1),axis.line.y = element_blank(), axis.text.x = element_blank(),axis.ticks.x = element_blank(),legend.title=element_blank())plot.histogram
legend = cowplot::get_legend(plot)
grid.newpage()
grid.draw(legend)

Exporting plot

output_name = 'PostSource.AcrossCases.Control.Strain'ggsave(plot + theme(legend.position = 'none'), device = 'pdf', dpi = 300, width = 3, height = 5,filename = paste0(results_dir, fig, output_name, '.Plot.pdf'))
ggsave(plot.histogram + theme(legend.position = 'none'), device = 'pdf', dpi = 300, width = 3, height = .6,filename = paste0(results_dir, fig, output_name, '.Histogram','.Plot.pdf'))
ggsave(legend, device = 'pdf', dpi = 300, width = 5, height = 2,filename = paste0(results_dir, fig, output_name, '.Legend.pdf'))

Strain-level Resolution

Format

resolution <- sstr_cases %>%filter(Study_Type == 'Control') %>% tag_species_competition(.) %>% mutate(n = 1) %>% mutate(source = ifelse(grepl(species, pattern = 'unclassified'),  'Unclassified', source)) %>% mutate(source = case_when(analysis_level == 'species' & source == 'self' ~ 'Initial Sample Sp.',analysis_level == 'species' & source == 'unique' ~ 'New Sp.',T ~ source)) %>% group_by(Study_Type, Case_Name, source, Days_Since_FMT.post, fmt_success.label) %>% summarize(n = sum(n, na.rm = T),rel_abund = sum(rel_abund.post, na.rm = T) / 100,.groups = 'drop') %>% group_by(Case_Name, Days_Since_FMT.post) %>% mutate(f = n / sum(n, na.rm = T)) %>%ungroup() %>% dplyr::select(-n) %>% group_by(Case_Name) %>% mutate(source = case_when(source == 'Unclassified' ~ 'Unclassified Sp.',source == 'same_species' ~ 'Same Sp.', source == 'unique' ~ 'New',source == 'self' ~ 'Initial Sample',T ~ source)) %>% pivot_wider(names_from = 'source', values_from = c('rel_abund','f'), names_sep = '___') %>% mutate_at(.vars = vars(contains('___')),.funs = funs(replace_na(., 0))) %>%pivot_longer(cols = contains("___"), names_to = c("metric", "source"),names_sep = '___', values_drop_na = F) %>% mutate(resolution = case_when(grepl(source, pattern = 'Sp.') ~ 'Species',T ~ 'Strain')) %>% mutate(metric = case_when(metric == 'rel_abund' ~ 'Rel.\nAbund.', metric == 'f' ~ 'Spp.\nFraction', T ~ metric)) %>% group_by(Case_Name, Days_Since_FMT.post, metric, resolution) %>% summarize(value = sum(value, na.rm = T), .groups = 'drop')

Relative abundance and fraction of species, on average, for which we obtained strain-level resolution

resolution %>% filter(resolution == 'Strain') %>% group_by(metric) %>% summarize(value.mean = mean(value, na.rm = T), value.sd = sd(value, na.rm = T), .groups = 'drop') 

Boxplots

plot.resolution <- resolution %>% filter(resolution == 'Strain') %>% ggplot(aes(metric, value)) + stat_boxplot(geom = 'errorbar') +geom_boxplot(show.legend = F) + geom_dotplot(aes(fill = NULL), binaxis = "y", stackdir = "center", binpositions="all", binwidth = 0.015) +scale_y_continuous(labels = percent_format(),breaks = seq(0, 1, .25),expand = c(0,0),limits = c(0, 1)) + theme_cowplot() +theme(legend.position = 'none', # c(.6, .65),legend.title = element_blank(),plot.subtitle = element_text(hjust = 0.5),axis.title = element_blank(),) + labs(subtitle = 'Strain-level Resolution')
plot.resolution

Exporting plot

output_name = 'PostSource.Control.Strain'ggsave(plot.resolution, device = 'pdf', dpi = 300, width = 2.25, height = 3.75,filename = paste0(results_dir, 'Fig_3.', output_name, '.ResolutionPlot.pdf'))

Shared Strains (rCDI)

Family Related

plot <-sstr.rcdi %>% filter(Pair_Type == 'Pre-FMT/Donor') %>%mutate(tag.facet = case_when(Related.label == 'Same Subject' ~ 'Same Individual', Related.label == 'Same Donor' & Pair_Type == 'Donor/Donor' ~ 'Same Individual',Related.label == 'Same Case' & Pair_Type != 'Donor/Post-FMT' ~ 'Same Individual',Related.label == 'Same Donor' & Pair_Type == 'Post-FMT/Post-FMT' ~ 'Treatment-related',Related.label %in% c('Same Donor', 'Same Case') & Pair_Type == 'Donor/Post-FMT' ~ 'Treatment-related',T ~ Related.label),tag.x = case_when(Related.label == 'Distinct Subject' ~ 'Unrelated\nSample Pair', T ~ gsub(Pair_Type, pattern = '/', replacement= '\n'))) %>% mutate(tag.x = case_when(Pair_Type == 'Pre-FMT/Donor' & Study.pair == 'FRICKE' ~'Family-related\nPre-FMT/Donor',T ~ tag.x),tag.x = case_when(Related.label == 'Same Donor' ~ paste0(gsub(tag.x, pattern = '\n', replacement ='*\n'), '*'),T ~ tag.x),tag.x = fct_relevel(as.factor(gsub(tag.x, pattern = '-FMT', replacement = '')), 'Pre\nPost', 'Donor\nPost')) %>% ggplot(aes(tag.x, n.shared_strains.mvs, fill = tag.facet)) + stat_boxplot(geom = 'errorbar') +geom_boxplot() + stat_compare_means(method = 'wilcox', label.x.npc = 0.5,exact = F, label = 'p.signif') +facet_grid(cols = vars(tag.facet),  scales = 'free', space = 'free_x') + scale_fill_manual(values = c('black', 'grey75','white')) +theme_cowplot() + guides(color = guide_legend(ncol = 1), linetype = guide_legend(ncol = 1)) + theme(legend.title=element_blank(),axis.title.x = element_blank(),panel.margin.y = unit(.5, "lines"),panel.background = element_blank(),strip.background = element_blank(), strip.text = element_blank(),axis.text = element_text(size = 9),axis.title = element_text(size = 11), axis.text.x = element_text(hjust = 0.5),text = element_text(colour = 'black'),) + labs(y = 'Strains Shared between Samples')plot + theme(legend.position = 'none')legend = cowplot::get_legend(plot)
grid.newpage()
grid.draw(legend)

FMT Related

plot <-sstr.rcdi %>% mutate(tag.facet = case_when(Related.label == 'Same Subject' ~ 'Same Individual', Related.label == 'Same Donor' & Pair_Type == 'Donor/Donor' ~ 'Same Individual',Related.label == 'Same Case' & Pair_Type != 'Donor/Post-FMT' ~ 'Same Individual',Related.label == 'Same Donor' & Pair_Type == 'Post-FMT/Post-FMT' ~ 'Same Donor',Related.label %in% c('Same Donor', 'Same Case') & Pair_Type == 'Donor/Post-FMT' ~ 'Same Donor',Related.label == 'Distinct Subject' ~ 'Different Individuals',T ~ 'Unknown'),tag.x = gsub(Pair_Type, pattern = '/', replacement = '\n'),tag.x = gsub(tag.x, pattern = '-FMT', replacement = ''),tag.x = ifelse(tag.x == 'Pre\nDonor', 'Donor\nPre', tag.x),tag.x = fct_relevel(tag.x,  'Pre\nPre', 'Pre\nPost', 'Post\nPost', 'Donor\nDonor', 'Donor\nPre', 'Donor\nPost')) %>% mutate(tag.facet = fct_relevel(tag.facet, 'Same Individual', 'Same Donor')) %>% mutate(tag.f = ifelse(grepl(tag.x, pattern = 'Donor'),  'Donor-Related', 'Patient-Related'),tag.f = fct_relevel(tag.f, 'Patient-Related')) %>% filter(!(tag.facet == 'Different Individuals' & tag.x %in% c('Pre\nPost', 'Donor\nPre', 'Donor\nPost'))) %>% ggplot(aes(tag.x, n.shared_strains.mvs, fill = tag.facet)) + stat_boxplot(geom = 'errorbar') +geom_boxplot() + facet_grid(cols = vars(tag.facet),scales = 'free', space = 'free_x') +scale_fill_manual(values = c('white', 'grey75','black')) +theme_cowplot() + guides(color = guide_legend(ncol = 1), linetype = guide_legend(ncol = 1)) + theme(legend.title=element_blank(),axis.title.x = element_blank(),panel.margin.y = unit(.5, "lines"),panel.background = element_blank(),strip.background = element_blank(), strip.text = element_blank(),axis.text = element_text(size = 9),axis.title = element_text(size = 11), axis.text.x = element_text(hjust = 0.5),text = element_text(colour = 'black'),) + labs(y = 'Strains Shared between Samples')plot + theme(legend.position = 'none')legend = cowplot::get_legend(plot)
grid.newpage()
grid.draw(legend)

Exporting plot

output_name = 'SharedStrains.rCDI'ggsave(plot + theme(legend.position = 'none'), device = 'pdf', dpi = 300, width = 4.75, height = 3.05,filename = paste0(results_dir, fig, output_name, '.Plot.pdf'))
ggsave(legend, device = 'pdf', dpi = 300, width = 2, height = 1,filename = paste0(results_dir, fig, output_name, '.Legend.pdf'))

Statistics

output_name = 'SharedStrains.rCDI'
sstr.rcdi %>% mutate(fmt_success.label = ifelse(fmt_success, 'Resolved', 'Failed')) %>% filter(Same_Case.label == 'Same Case') %>% dplyr::select(Study.pair, Case_Name.pair, Pair_Type, fmt_success.label, starts_with('n.shared'), starts_with('Days_Since_FMT')) %>% filter(Pair_Type %in% c('Pre-FMT/Post-FMT', 'Donor/Post-FMT')) %>% rowwise() %>% mutate(Days_Since_FMT = max(c(Days_Since_FMT.row, Days_Since_FMT.col))) %>% ungroup() %>% group_by(Case_Name.pair) %>% filter(dense_rank(Days_Since_FMT) == max(dense_rank(Days_Since_FMT))) %>% ungroup() %>% dplyr::select(-Days_Since_FMT.row, -Days_Since_FMT.col, -n.shared_strains.cvs, -n.shared_genus, -n.shared_family) %>% pivot_wider(names_from = 'Pair_Type', values_from = starts_with('n.shared'), id_cols = c('Days_Since_FMT', 'Study.pair', 'Case_Name.pair', 'fmt_success.label')) %>% dplyr::select(Study.pair, Case_Name.pair, fmt_success.label, Days_Since_FMT, starts_with('n.shared_strains'), starts_with('n.shared_species')) %>% write_tsv(., paste0(results_dir, fig, output_name, '.Counts.tsv'))

Labelling Control

Format

library(tidygraph)
library(ggraph)

sstr_cooccurrence <- read_tsv(file = paste0(data_dir, 'samples.sstr_cooccurrence.tsv'))

Set Figure

fig = paste0('Suppl.Fig_6.')

get_matrix <- function(data, selection, min_strain=3) {sstr_select <-data %>%filter(row %in% selection,col %in% selection)sstr_matrix <-bind_rows(sstr_select, sstr_select %>% rename(row = col, col = row)) %>% pivot_wider(names_from = col, values_from = n.shared_strain,values_fill = list(n.shared_strain = 0)) %>% column_to_rownames('row')idx <- intersect(rownames(sstr_matrix), colnames(sstr_matrix))sstr_matrix <- sstr_matrix[idx, idx]sstr_matrix[as.matrix(sstr_matrix) <= min_strain] <- 0return(sstr_matrix)
}get_graph <- function(sstr_matrix) {sstr_graph <- as_tbl_graph(as.matrix(sstr_matrix), directed = F) %>% activate(nodes) %>%left_join(., samples.metadata, by = c('name' = 'Name'))return(sstr_graph)
}generate_graph <- function(data, selection, min_strain=3) {return(get_graph(get_matrix(data, selection, min_strain=3)))
}

LOUISS

LOUISS Case 64 / 66 last time point mislabelling

selection <- samples.metadata %>% filter(Case_Name %in% c('LOUISS_Case_64', 'LOUISS_Case_66')) %>% pull(Name)misl1 <- get_matrix(sstr_cooccurrence, selection, 0) %>% rownames_to_column('row') %>% pivot_longer(names_to = 'col', values_to = 'shared',cols = ends_with('.pair')) %>% filter(row != col) %>% separate(row, into = c('Study.row','Case_Name.row','Time_point.row'), sep = '_', remove = F) %>% separate(col, into = c('Study.col','Case_Name.col','Time_point.col'), sep = '_', remove = F) %>% filter(!col %in% c('LOUISS_AS64_24.pair','LOUISS_AS66_24.pair')) %>% mutate(Same = Case_Name.col == Case_Name.row,Name = str_remove(row, pattern = '.pair')) %>% dplyr::select(Name, shared, Same) %>% group_by(Name, Same) %>% summarize_all(.funs = funs(mean, sd)) %>% mutate(Same = ifelse(Same, 'Original Case', 'Other Case')) %>% pivot_wider(names_from = 'Same', names_sep = '.',values_from = c('mean','sd')) %>% dplyr::select(Name, contains('Original'), everything()) %>% separate(Name, into = c('Study','Case Name','Time point')) %>% arrange(`Case Name`, as.numeric(`Time point`)) %>% mutate_at(.vars = vars(starts_with('mean'), starts_with('sd')),.funs = funs(round(., 2)))

Denoising up to 3 shared strains, showing all connections above 3 strains

selection <- samples.metadata %>% filter(grepl(Study, pattern = 'LOUISS')) %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)yy = -.6
c.col = get_palette(c('black','white'), 15)plot <- ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 3, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 5, aes(fill = str_remove(Case_Name, pattern = 'LOUISS_Case_')),) + geom_node_label(aes(label = Sample, filter = name %in% c('LOUISS_AS66_24.pair', 'LOUISS_AS64_24.pair')), repel = T, hjust = .25, vjust = -3) +scale_y_continuous(expand = c(0.6, 0.6)) +scale_x_reverse(expand = c(0.1, 0.1)) +theme_graph() +theme(legend.position = 'none', strip.background = element_blank(),panel.spacing.y = unit(0, 'cm'),strip.placement = 'inside',strip.text = element_blank()) + guides(edge_alpha = F, fill = guide_legend(override.aes=list(shape=21), title = '', nrow = 2)) + scale_edge_alpha_continuous(breaks = c(5, 10, 25, 50)) +scale_fill_manual(values = c(# 43c.col[1],# 44c.col[5],# 45c.col[2],# 50c.col[6],# 51c.col[7],# 53c.col[8],# 56c.col[9],# 58c.col[10],# 60c.col[11],# 61c.col[3],# 62c.col[12],# 63c.col[13],# 64colors.discrete[4],# 65c.col[14],# 66'#B4907E',# 68c.col[4])) + annotate('text', x = .6, y = yy, label = '43') + annotate('text', x = 2.75, y = yy, label = '45') + annotate('text', x = 4.75, y = yy, label = '61') + annotate('text', x = 7, y = yy, label = '68') + annotate('text', x = 9.5, y = yy, label = '44') + annotate('text', x = 11.75, y = yy, label = '50') + annotate('text', x = 13.75, y = yy, label = '51') + annotate('text', x = 16.25, y = yy, label = '53') + annotate('text', x = 18.35, y = yy, label = '56') + annotate('text', x = 20.6, y = yy, label = '58') + annotate('text', x = 23, y = yy, label = '60') + annotate('text', x = 25.25, y = yy, label = '62') + annotate('text', x = 27.5, y = yy, label = '63') + annotate('text', x = 29.75, y = yy, label = '64') + annotate('text', x = 32, y = yy, label = '66') + annotate('text', x = 34.25, y = yy, label = '65') + facet_wrap(~ x < 20, scales = 'free', nrow = 2)
plot

Exporting plot

output_name = 'Mislabelling.Example1'ggsave(plot, device = 'pdf', dpi = 300, width = 8, height = 5,filename = paste0(results_dir, fig, output_name, '.Plot.pdf'))
misl1 %>% write_tsv(., paste0(results_dir, fig, output_name, '.tsv'))

ALM

ALM patient sample with many Donor strains BEFORE FMT labelled as recipient / pre-FMT
Collected on day of FMT - As per C. Smillie not clear if before or after from notes

selection <- samples.metadata %>% filter(Donor.Subject == 'D0_ALM_Case_4;5;6;8;9;12;13;15;17;18;20;21') %>% pull(Name)samples.metadata.orig <-samples.metadata %>% dplyr::select(Name, Case_Name, Sample_Type) %>% mutate(Sample_Type = ifelse(Name == 'ALM_FMT15.pair', 'recipient', Sample_Type))misl2 <- get_matrix(sstr_cooccurrence, selection, 0) %>% rownames_to_column('row') %>% pivot_longer(names_to = 'col', values_to = 'shared',cols = ends_with('.pair')) %>% filter(row != col) %>% left_join(., samples.metadata.orig, by = c('row' = 'Name')) %>% left_join(., samples.metadata.orig, by = c('col' = 'Name'),suffix = c('.row','.col')) %>% filter(Sample_Type.row == 'recipient', Sample_Type.col != 'recipient') %>% mutate(Same = case_when(Sample_Type.col == 'post' & Case_Name.col == Case_Name.row ~ 'Self Post-FMT',Sample_Type.col == 'post' & Case_Name.col != Case_Name.row ~ 'Other Post-FMT',Sample_Type.col == 'donor' ~ 'Donor',)) %>% mutate(Recipient = str_remove(row, pattern = '.pair'),Study = 'ALM') %>% rename(Case_Name = 'Case_Name.row') %>% dplyr::select(Study, Case_Name, Recipient, shared, Same) %>% group_by(Study, Case_Name, Recipient, Same) %>% summarize_all(.funs = funs(mean, sd)) %>% pivot_wider(names_from = 'Same', names_sep = '.',values_from = c('mean','sd')) %>% dplyr::select(Study, Case_Name, Recipient, contains('Self'), contains('Other'), everything()) %>% mutate_at(.vars = vars(starts_with('mean'), starts_with('sd')),.funs = funs(replace_na(round(., 2), 0))) %>% mutate_at(.vars = vars(Case_Name, Recipient),.funs = funs(str_remove(., 'ALM_'))) %>% arrange(Recipient)

Denoising below 3 shared strains, showing all connections >4 strains

selection <- samples.metadata %>% filter(Donor.Subject == 'D0_ALM_Case_4;5;6;8;9;12;13;15;17;18;20;21') %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)plot <-ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 5, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 8, aes(fill = case_when(Days_Since_FMT == 0.5 ~ 'Recipient', Sample_Type == 'post' ~ 'Post-FMT', Sample_Type == 'donor' ~ 'Donor', Sample_Type == 'recipient' ~ 'Recipient', T ~ '?'))) + geom_node_text(size = 3, aes(label = ifelse(Sample_Type == 'donor', 'D', Case_Number),col = ifelse(Sample_Type == 'recipient' | name == 'ALM_FMT15.pair', 'R', 'DP')),) + geom_node_label(aes(label = Sample, filter = name %in% c('ALM_FMT15.pair')), repel = T, hjust = .75, vjust = -8) +scale_y_continuous(expand = c(0.25, 0)) + theme_graph(base_family = 'Helvetica') + # theme(legend.key.width = unit(1, 'cm'),# legend.key.height = unit(0.5, 'cm')) + scale_fill_manual(values = colors.discrete[c(1,2,3)]) + scale_edge_alpha_continuous(breaks = c(5, 10, 25, 50)) +scale_color_manual(values = c('black', 'white')) + guides(edge_alpha = guide_legend(title = 'Shared\nStrains',override.aes = list(size = 5)), col = F, fill = F)
plot + theme(legend.position = 'none')
legend = cowplot::get_legend(plot)
grid.newpage()
grid.draw(legend)

Exporting plot

output_name = 'Mislabelling.Example2'ggsave(plot + theme(legend.position = 'none'), device = 'pdf', dpi = 300, width = 11, height = 8,filename = paste0(results_dir, fig, output_name, '.Plot.pdf'))
ggsave(legend, device = 'pdf', dpi = 300, width = 1, height = 1.5,filename = paste0(results_dir, fig, output_name, '.Legend.pdf'))misl2 %>% write_tsv(., paste0(results_dir, fig, output_name, '.tsv'))

selection <- samples.metadata %>% filter(Donor.Subject == 'D0_ALM_Case_22;23;24;25') %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 3, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 8, aes(fill = case_when(Days_Since_FMT == 0.5 ~ 'Recipient', Sample_Type == 'post' ~ 'Post-FMT', Sample_Type == 'donor' ~ 'Donor', Sample_Type == 'recipient' ~ 'Recipient', T ~ '?'))) + geom_node_text(col = 'black',size = 3.5, aes(label = ifelse(Sample_Type == 'donor', 'D', Case_Number))) + geom_node_label(aes(label = Sample, filter = name %in% c('ALM_FMT15.pair')), repel = T, hjust = .75, vjust = -3) +scale_y_continuous(expand = c(0.25, 0)) + scale_x_continuous(expand = c(0.25, 0.25)) + theme_graph() + guides(edge_alpha = F, fill = guide_legend(override.aes = list(shape=21), title = 'Smillie et al.\noriginal label', ncol = 1))

selection <- samples.metadata %>% filter(Donor.Subject == 'D0_ALM_Case_1;3') %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 3, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 8, aes(fill = case_when(Days_Since_FMT == 0.5 ~ 'Recipient', Sample_Type == 'post' ~ 'Post-FMT', Sample_Type == 'donor' ~ 'Donor', Sample_Type == 'recipient' ~ 'Recipient', T ~ '?'))) + geom_node_text(col = 'black',size = 3.5, aes(label = ifelse(Sample_Type == 'donor', 'D', Case_Number))) + geom_node_label(aes(label = Sample, filter = name %in% c('ALM_FMT15.pair')), repel = T, hjust = .75, vjust = -5) +scale_y_continuous(expand = c(0.25, 0)) + scale_x_continuous(expand = c(0.25, 0.25)) +  theme_graph() + guides(edge_alpha = F, fill = guide_legend(override.aes = list(shape=21), title = 'Smillie et al.\noriginal label', ncol = 1))

FRICKE

selection <- samples.metadata %>% filter(Study == 'FRICKE') %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 3, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 8, aes(fill = case_when(Days_Since_FMT == 0.5 ~ 'Recipient', Sample_Type == 'post' ~ 'Post-FMT', Sample_Type == 'donor' ~ 'Donor', Sample_Type == 'recipient' ~ 'Recipient', T ~ '?'))) + geom_node_text(col = 'black',size = 3.5, aes(label = Case_Number)) + geom_node_label(aes(label = Sample, filter = name %in% c('ALM_FMT15.pair')), repel = T, hjust = .75, vjust = -5) +scale_y_continuous(expand = c(0.25, 0)) + theme_graph() + guides(edge_alpha = F, fill = guide_legend(override.aes = list(shape=21), title = 'Sample Type', ncol = 1))

RAYMONDF

selection <- samples.metadata %>% filter(Study == 'RAYMONDF') %>% pull(Name)sstr_graph <- generate_graph(sstr_cooccurrence, selection, 3)ggraph(sstr_graph, layout = 'stress') +geom_edge_link(aes(filter = weight >= 3, alpha = weight)) +geom_node_point(shape = 21, col = 'black',size = 8, aes(fill = Days_Since_FMT)) + geom_node_text(col = 'black',size = 3.5, aes(label = Case_Number)) + geom_node_label(aes(label = Sample, filter = name %in% c('ALM_FMT15.pair')), repel = T, hjust = .75, vjust = -5) +scale_y_continuous(expand = c(0.25, 0)) + theme_graph() + guides(edge_alpha = F, fill = guide_legend(override.aes = list(shape=21), title = 'Sampling Day', ncol = 1))