Analyzing monoculture growth summary

1 Introduction

In the prior step we imported, smoothed, and calculated summary statistics for the species growth in monocultures. These monocultures include experiments either with 100% R2A and different levels of streptomycin or with no streptomycin and the pairwise filtrates from all other species.

Here we will do some simple plots and analysis of the growth data. We will also look into the filtrate data.

2 Setup

2.1 Libraries

library(tidyverse)
library(here)
library(fs)
library(scales)
library(ggraph)
source(here::here("R", "utils_generic.R"))

2.2 Global variables

data_raw <- here::here("_data_raw", "monocultures", "20240328_bioscreen")
data <- here::here("data", "monocultures", "20240328_bioscreen")

# make processed data directory if it doesn't exist
fs::dir_create(data)

2.3 Read growth summary data

many_auc_res <- readr::read_tsv(here::here(data, "gcurve_auc_results.tsv"))
many_spline_res <- readr::read_tsv(here::here(data, "gcurve_spline_results.tsv"))
many_linear_res <- readr::read_tsv(here::here(data, "gcurve_linear_results.tsv"))
many_baranyi_res <- readr::read_tsv(here::here(data, "gcurve_baranyi_results.tsv"))
many_huang_res <- readr::read_tsv(here::here(data, "gcurve_huang_results.tsv"))

3 Select and format

The past notebook showed that different nonparametric and parametric models fit the data with different degrees of quality. Overall, it appears that the Huang parametric model fits the data the best so we will default to using the Huang growth rates. If the Huang doesn’t fit well

Moving forward we’ll use Huang-derived growth rates if the fit looks good. Otherwise we will go with the spline estimate which seemed to perform pretty well and was quite flexible.

Here we determine a threshold for filtering out cultures that don’t grow as 1.25\(\times\) the mean minimum optical density of all cultures. For these cultures growth rates estimates are not really reliable.

many_auc_res %>% 
  summarize(mn = mean(min_od), 
            sd = sd(min_od)) %>% 
  mutate(thresh = 1.25*mn) %>% 
  pull(thresh)

[1] 0.1019044

best_parametric <- bind_rows(
  dplyr::select(many_huang_res, plate_name:sp_filtrate, y0, mumax, K, r2) %>% mutate(model = "huang"),
  dplyr::select(many_baranyi_res, plate_name:sp_filtrate, y0, mumax, K, r2) %>% mutate(model = "baranyi")) %>% 
  arrange(plate_name, bioscreen_well) %>% 
  group_by(plate_name, bioscreen_well) %>% 
  filter(r2 == max(r2)) %>% 
  ungroup()

The Huang model fits best for about 75% of all growth curves.

best_parametric %>% 
  count(model) %>% 
  mutate(p = n/sum(n))

4 Update!!

After looking at this more and more I think for the rest of this analysis we will just go with the spline non-parametric fits. Sometimes the Huang model fits really really well, but other times it doesn’t. It’s not worth going into every single growth curve to find out

5 Monoculture growth with streptomycin

We’ll now examine how the monoculture growth changed with respect to evolutionary history and streptomycin concentrations

5.1 Growth rates

gr_strep <- many_spline_res %>% 
  filter(!str_detect(plate_name, "pairwise")) %>% 
  mutate(hist = str_split_i(sp_hist, "_", 1),
         sp = str_split_i(sp_hist, "_", 2)) %>% 
  summarize(ggplot2::mean_cl_boot(mumax), .by=c(sp, hist, strep_conc)) %>% 
  ggplot(aes(x = strep_conc, y = y)) +
  geom_linerange(aes(ymin = ymin, ymax = ymax, color = hist)) +
  geom_line(aes(color = hist), lty=2) +
  geom_point(aes(color = hist)) +
  labs(y = "Maximum per capita growth rate μ (hr-1)", x = "Streptomycin conc. (μg/ml)", 
       color = "Evolutionary\nhistory") +
  facet_grid(~sp) +
  scale_x_continuous(trans="log1p", breaks = c(0, 1, 10, 100, 1000, 5000))

ggsave(
  here::here("figs", "monoculture_gr_strep.svg"),
  gr_strep,
  width = 10,
  height = 4,
  units = "in",
  device = "svg"
)

ggsave(
  here::here("figs", "monoculture_gr_strep.png"),
  gr_strep,
  width = 10,
  height = 4,
  units = "in",
  device = "png"
)

Figure 1: Maximum specific growth rate (\(\mathrm{\mu_{max}}\) in \(\mathrm{hr^{-1}}\)) for the four different bacterial species (panels) and their different evolutionary histories (colors) at different streptomycin concentrations (horizontal axis, μg/ml). Points are the mean over five biological replicates and line ranges show the 95% confidence interval from bootstraps. EVO = experimentally evolved to high streptomycin.

5.2 AUC (area under the growth curve)

auc_strep <- many_auc_res %>% 
  filter(!str_detect(plate_name, "pairwise")) %>% 
  mutate(hist = str_split_i(sp_hist, "_", 1),
         sp = str_split_i(sp_hist, "_", 2)) %>% 
  summarize(ggplot2::mean_cl_boot(auc), .by=c(sp, hist, strep_conc)) %>% 
  ggplot(aes(x = strep_conc, y = y)) +
  geom_linerange(aes(ymin = ymin, ymax = ymax, color = hist)) +
  geom_line(aes(color = hist), lty=2) +
  geom_point(aes(color = hist)) +
  labs(y = "Total area under the growth curve", x = "Streptomycin conc. (μg/ml)", 
       color = "Evolutionary\nhistory") +
  facet_grid(~sp) +
  scale_x_continuous(trans="log1p", breaks = c(0, 1, 10, 100, 1000, 5000))

ggsave(
  here::here("figs", "monoculture_auc_strep.svg"),
  auc_strep,
  width = 10,
  height = 4,
  units = "in",
  device = "svg"
)

ggsave(
  here::here("figs", "monoculture_auc_strep.png"),
  auc_strep,
  width = 10,
  height = 4,
  units = "in",
  device = "png"
)

Figure 2: Area under the growth curve for the four different bacterial species (panels) and their different evolutionary histories (colors) at different streptomycin concentrations (horizontal axis, μg/ml). Points are the mean over five biological replicates and line ranges show the 95% confidence interval from bootstraps. EVO = experimentally evolved to high streptomycin.

6 Pairwise filtrate growth

This is information about how well each species grew on the filtrate of all the species.

many_auc_res

a <- many_spline_res %>% 
  filter(strep_conc == 0) %>% 
  summarize(m_max = mean(mumax), .by=c(sp_hist))

many_spline_res %>% 
  filter(str_detect(plate_name, "pairwise")) %>% 
  summarize(m = mean(mumax), .by=c(sp_hist, sp_filtrate)) %>% 
  left_join(a, by = join_by(sp_hist)) %>% 
  mutate(m_rel = m/m_max*100)

plot_heatmap <- function(df, fill_var, fill_lab){
  ggplot(df, aes(x = sp_filtrate, y = sp_hist)) +
  geom_tile(aes(fill = {{fill_var}})) +
  geom_text(aes(label = paste0(round({{fill_var}}), "%"), color = {{fill_var}} > 40)) +
  scale_x_discrete(guide = guide_axis(angle = 45)) +
  scale_y_discrete(guide = guide_axis(angle = 45)) +
  scale_fill_viridis() +
  scale_color_manual(values = c("white", "black"), guide = "none") +
  labs(y = "Growth of species/evolutionary history", x = "Filtrate from species/evolutionary history",
       fill = fill_lab) +
  coord_fixed() + 
  ggplot2::theme(panel.grid = element_blank(),
                 panel.background = element_blank(), 
                 strip.background = element_blank(),
                 panel.border = element_blank())
}

6.1 Growth rate

gr_max <- many_spline_res %>% 
  filter(strep_conc == 0) %>% 
  summarize(m_max = mean(mumax), .by=c(sp_hist))

gr_filtrate <- many_spline_res %>% 
  filter(str_detect(plate_name, "pairwise")) %>% 
  summarize(m_filt = mean(mumax), .by=c(sp_hist, sp_filtrate)) %>% 
  left_join(gr_max, by = join_by(sp_hist)) %>% 
  mutate(m_rel = m_filt/m_max*100) %>% 
  mutate(sp_filtrate = factor(sp_filtrate, levels =c("ANC_0403", "EVO_0403", "ANC_1287", "EVO_1287", "ANC_1896", 
                                              "EVO_1896", "ANC_1977", "EVO_1977")),
         sp_hist = factor(sp_hist, levels = c("ANC_0403", "EVO_0403", "ANC_1287", "EVO_1287", "ANC_1896", 
                                              "EVO_1896", "ANC_1977", "EVO_1977")))

gr_filtrate_plot <- plot_heatmap(gr_filtrate, m_rel, "% of max μ")

ggsave(
  here::here("figs", "monoculture_gr_filtrate.svg"),
  gr_filtrate,
  width = 8,
  height = 8,
  units = "in",
  device = "svg"
)

Figure 3: Percent maximum growth rate (μ) of each species/evolutionary history (vertical axis) growing on the filtrate of other species, including itself (horizontal axis). Maximum μ is from monocultures with no competitor or streptomycin. Filtrate μ is from each species grown on the filtrate of every other species.

6.2 AUC (area under the growth curve)

auc_max <- many_auc_res %>% 
  filter(strep_conc == 0) %>% 
  summarize(auc_max = mean(auc), .by=c(sp_hist))

auc_filtrate <- many_auc_res %>% 
  filter(str_detect(plate_name, "pairwise")) %>% 
  summarize(auc_filt = mean(auc), .by=c(sp_hist, sp_filtrate)) %>% 
  left_join(auc_max, by = join_by(sp_hist)) %>% 
  mutate(auc_rel = auc_filt/auc_max*100) %>% 
  mutate(sp_filtrate = factor(sp_filtrate, levels =c("ANC_0403", "EVO_0403", "ANC_1287", "EVO_1287", "ANC_1896", 
                                              "EVO_1896", "ANC_1977", "EVO_1977")),
         sp_hist = factor(sp_hist, levels = c("ANC_0403", "EVO_0403", "ANC_1287", "EVO_1287", "ANC_1896", 
                                              "EVO_1896", "ANC_1977", "EVO_1977")))

auc_filtrate_plot <- plot_heatmap(auc_filtrate, auc_rel, "% of max AUC")

ggsave(
  here::here("figs", "monoculture_auc_filtrate.svg"),
  auc_filtrate,
  width = 8,
  height = 8,
  units = "in",
  device = "svg"
)

Figure 4: Percent maximum Area Under the Growth Curve (AUC) of each species/evolutionary history (vertical axis) growing on the filtrate of other species, including itself (horizontal axis). Maximum AUC is from monocultures with no competitor or streptomycin. Filtrate AUC is from each species grown on the filtrate of every other species.