In WendyLiuLab/BiactivationFlow: What the package does (short line)
library(dplyr)
library(ggplot2)
library(reshape2)
library(testthat)
library(FlowAnalysis)
protocols = c("simultaneous", "m1_sus_m2", "m2_sus_m1") # m1_m2, m2_m1
to_save = NULL
for(stub in protocols) {
protocol = eval(parse(text=stub))
normalized = normalize_by_positive_controls(protocol)
normalized_medians = as.data.frame(make_medians(normalized))
normalized_medians_mean = normalized_medians %>%
group_by(antibody, m1_concentration, m2_concentration) %>%
summarize(mean.CD86=mean(CD86), sd.CD86=sd(CD86),
mean.CD206=mean(CD206), sd.CD206=sd(CD206),
n=n()) %>%
mutate(sem.CD86=sd.CD86/sqrt(n), sem.CD206=sd.CD206/sqrt(n)) %>%
mutate(ymax.CD86=mean.CD86+sem.CD86, ymin.CD86=mean.CD86-sem.CD86,
ymax.CD206=mean.CD206+sem.CD206, ymin.CD206=mean.CD206-sem.CD206)
eval(parse(text=sprintf("normalized_%s = normalized", stub)))
eval(parse(text=sprintf("normalized_%s_medians = normalized_medians", stub)))
eval(parse(text=sprintf("normalized_%s_medians_mean = normalized_medians_mean", stub)))
to_save = c(to_save,
# sprintf("normalized_%s", stub),
sprintf("normalized_%s_medians", stub),
sprintf("normalized_%s_medians_mean", stub))
}
# simul_timecourse is special
normalized_simul_timecourse_medians = simul_timecourse %>%
normalize_by_positive_controls_timepoint %>%
group_by(Experiment, antibody, timepoint, m1_concentration, m2_concentration) %>%
summarize(CD206 = median(CD206), CD86 = median(CD86))
normalized_simul_timecourse_medians_mean = normalized_simul_timecourse_medians %>%
group_by(m1_concentration, m2_concentration, timepoint, antibody) %>%
summarize(mean.CD86=mean(CD86), sd.CD86=sd(CD86),
mean.CD206=mean(CD206), sd.CD206=sd(CD206),
n=n()) %>%
mutate(sem.CD86=sd.CD86/sqrt(n), sem.CD206=sd.CD206/sqrt(n)) %>%
mutate(ymax.CD86=mean.CD86+sem.CD86, ymin.CD86=mean.CD86-sem.CD86,
ymax.CD206=mean.CD206+sem.CD206, ymin.CD206=mean.CD206-sem.CD206,
M1xM2=paste(m1_concentration, m2_concentration, sep="x"))
manuscript_theme = theme_bw() + theme(text=element_text(size=8))
Figure 1
1a. M1 dose-response histogram
g = simultaneous %>%
filter(Experiment == "tds-004-03", m2_concentration == 0) %>%
mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>%
ggplot(aes(CD86, color=antibody)) +
facet_grid(m1_concentration~.) +
geom_density() +
scale_color_discrete("Antibody") +
scale_x_continuous(trans=biexp_trans(lim=100, decade.size=400),
breaks=c(-100, 0,100,1000,10000,100000),
labels=c("-100", "0","100","1000","1e4","1e5")) +
scale_y_continuous(limits=c(0, 0.006),
breaks=c(0, 0.003, 0.006)) +
labs(x="CD86 intensity", y="Density") +
geom_vline(
data=(simultaneous %>%
filter(Experiment == "tds-004-03", m2_concentration == 0) %>%
mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>%
group_by(Experiment, m1_concentration, antibody) %>%
summarize(median.CD86=median(CD86))),
mapping=aes(xintercept=median.CD86, color=antibody),
alpha=0.5
) +
guides(color=guide_legend(direction="horizontal")) +
manuscript_theme +
theme(legend.position="bottom")
print(g)
1b. M1 response increases with M1 stimulus
plot_1_filter = function(x) filter(x, antibody == "exp", m2_concentration == 0)
g = normalized_simultaneous_medians %>%
plot_1_filter %>%
ggplot(aes(m1_concentration)) +
geom_point(aes(y=CD86), alpha=0.2) +
geom_line(aes(y=CD86, group=Experiment), alpha=0.2) +
geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean),
aes(ymax=ymax.CD86, ymin=ymin.CD86),
width=0.2) +
geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean),
aes(ymax=mean.CD86, ymin=mean.CD86),
width=0.15) +
scale_x_discrete("[LPS/IFN-γ] (ng/ml)") +
scale_y_continuous("CD86 intensity relative to M1",
limits=c(0, 1.1),
breaks=seq(0, 1, 0.2)) +
manuscript_theme +
annotate(geom="text", x=c(3,4)+.15, y=c(.67,1.01), label="*", size=6)
print(g)
plot_1_filter = function(x) filter(x, antibody == "exp", m2_concentration == 0)
g = normalized_simultaneous_medians %>%
plot_1_filter %>%
ggplot(aes(m1_concentration)) +
geom_point(aes(y=CD86), alpha=0.2) +
geom_line(aes(y=CD86, group=Experiment), alpha=0.2) +
geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean),
aes(ymax=ymax.CD86, ymin=ymin.CD86),
width=0.2) +
geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean),
aes(ymax=mean.CD86, ymin=mean.CD86),
width=0.15) +
scale_x_discrete("[LPS/IFN-γ] (ng/ml)") +
scale_y_continuous("CD86 intensity relative to M1",
limits=c(0, 1.1),
breaks=seq(0, 1, 0.2)) +
manuscript_theme +
annotate(geom="text", x=c(3,4)+.15, y=c(.67,1.01), label="*", size=6)
print(g)
Post-hoc pairwise significance:
p1f = plot_1_filter(normalized_simultaneous_medians)
plot_1_anova = aov(CD86~m1_concentration, p1f)
print(summary(plot_1_anova))
print(pairwise.t.test(p1f$CD86, p1f$m1_concentration, "holm"))
Linear regression with dummy coding (significance means different from no treatment):
print(summary(lm(CD86~m1_concentration, p1f)))
1c. M2 dose-response histogram
g = simultaneous %>%
filter(Experiment == "tds-004-03", m1_concentration == 0) %>%
mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>%
ggplot(aes(CD206, color=antibody)) +
facet_grid(m2_concentration~.) +
geom_density() +
scale_color_discrete("Antibody") +
scale_x_continuous(trans=biexp_trans(lim=100, decade.size=400),
breaks=c(-100, 0,100,1000,10000,100000),
labels=c("-100", "0","100","1000","1e4","1e5"),
limits=c(-100,NA)) +
scale_y_continuous(limits=c(0, 0.006),
breaks=c(0, 0.003, 0.006)) +
labs(x="CD206 intensity", y="Density") +
geom_vline(
data=(simultaneous %>%
filter(Experiment == "tds-004-03", m1_concentration == 0) %>%
mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>%
group_by(Experiment, m2_concentration, antibody) %>%
summarize(median.CD206=median(CD206))),
mapping=aes(xintercept=median.CD206, color=antibody),
alpha=0.5
) +
guides(color=guide_legend(direction="horizontal")) +
manuscript_theme +
theme(legend.position="bottom")
print(g)
How much does M1 response increase over baseline?
f1_baseline = normalized_simultaneous_medians %>%
filter(antibody == "exp", m1_concentration==0, m2_concentration==0)
f1_cd86_t = t.test(f1_baseline$CD86, mu=1)
print(1/f1_cd86_t$estimate)
print(1/f1_cd86_t$conf.int)
1d. M2 response increases with M2 stimulus
plot_2_filter = function(x) filter(x, antibody == "exp", m1_concentration == 0)
g = normalized_simultaneous_medians %>%
plot_2_filter %>%
ggplot(aes(m2_concentration)) +
geom_point(aes(y=CD206), alpha=0.2) +
geom_line(aes(y=CD206, group=Experiment), alpha=0.2) +
geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean),
aes(ymax=ymax.CD206, ymin=ymin.CD206),
width=0.2) +
geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean),
aes(ymax=mean.CD206, ymin=mean.CD206),
width=0.15) +
scale_x_discrete("[IL-4/IL-13] (ng/ml)") +
scale_y_continuous("CD206 intensity relative to M2",
limits=c(0, 1.1),
breaks=seq(0, 1, 0.2)) +
manuscript_theme +
annotate(geom="text", x=c(3.05, 4)+.12, y=c(.59, 1.01), label="*", size=6)
print(g)
plot_2_filter = function(x) filter(x, antibody == "exp", m1_concentration == 0)
g = normalized_simultaneous_medians %>%
plot_2_filter %>%
ggplot(aes(m2_concentration)) +
geom_point(aes(y=CD206), alpha=0.2) +
geom_line(aes(y=CD206, group=Experiment), alpha=0.2) +
geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean),
aes(ymax=ymax.CD206, ymin=ymin.CD206),
width=0.2) +
geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean),
aes(ymax=mean.CD206, ymin=mean.CD206),
width=0.15) +
scale_x_discrete("[IL-4/IL-13] (ng/ml)") +
scale_y_continuous("CD206 intensity relative to M2",
limits=c(0, 1.1),
breaks=seq(0, 1, 0.2)) +
manuscript_theme +
annotate(geom="text", x=c(3.05, 4)+.12, y=c(.59, 1.01), label="*", size=6)
print(g)
Post-hoc pairwise significance:
p2f = plot_2_filter(normalized_simultaneous_medians)
plot_2_anova = aov(CD206~m2_concentration, p2f)
print(summary(plot_2_anova))
print(pairwise.t.test(p2f$CD206, p2f$m2_concentration, "holm"))
Linear regression with dummy coding (significance means different from no treatment):
print(summary(lm(CD206~m2_concentration, p2f)))
How much does M2 response increase over baseline?
f1_cd206_t = t.test(f1_baseline$CD206, mu=1)
print(1/f1_cd206_t$estimate)
print(1/f1_cd206_t$conf.int)
Figure 2
Figure 2b
g = normalized_simultaneous %>%
filter(antibody == "exp", Experiment == "tds-003-94") %>%
ggplot(aes(log2(CD86), log2(CD206))) +
stat_density2d(aes(alpha=..level..), geom="polygon", n=256) +
# xlim(-1, NA) + # ylim(-1, NA) +
# scale_y_continuous(breaks=c(0, 2.5, 5), limits=c(-1, NA)) +
xlim(-10,5) + ylim(-4,6) +
geom_hline(yintercept=0, alpha=0.5) + geom_vline(xintercept=0, alpha=0.5) +
coord_cartesian(xlim=c(-6,6), ylim=c(-4,4)) +
facet_grid(m2_concentration~m1_concentration) +
scale_alpha_continuous("Density", trans="sqrt", guide="none") +
labs(x="CD86 intensity relative to M1", y="CD206 intensity relative to M2") +
manuscript_theme +
geom_point(data=filter(normalized_simultaneous_medians, Experiment=="tds-003-94", antibody=="exp"),
mapping=aes(log2(CD86), log2(CD206)), color="red", size=0.6)
print(g)
Figure 2c, 2d
p14.cd86.ann = data.frame(
m1_concentration=c("0.1", "0.3", "0.3"),
x=c(1, 1, 2),
xend=c(4, 2, 4),
y=c(2, 1.9, 2.2)
)
(normalized_simultaneous_medians_mean %>% filter(antibody == "exp") %>%
ggplot(aes(m2_concentration, mean.CD86, group=m1_concentration)) +
facet_grid(m1_concentration~.) +
geom_point(size=0.6) +
geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) +
geom_path(alpha=0.3) +
ylim(0, 2.5) +
# ggtitle("CD86 vs M2 dose, by M1 dose") +
labs(x="[IL-4/IL-13] (ng/ml)", y="Mean CD86 intensity relative to M1") +
manuscript_theme +
geom_segment(data=p14.cd86.ann, mapping=aes(x=x, xend=xend, y=y, yend=y), show.legend=FALSE) +
geom_text(data=p14.cd86.ann, mapping=aes(x=(x+xend)/2, y=y+0.08), label="*", show.legend=FALSE, size=4) +
geom_segment(data=p14.cd86.ann, mapping=aes(x=x, xend=x, y=y, yend=y-0.1), show.legend=FALSE) +
geom_segment(data=p14.cd86.ann, mapping=aes(x=xend, xend=xend, y=y, yend=y-0.1), show.legend=FALSE)
) %>% print
normalized_simultaneous_medians = within(normalized_simultaneous_medians, {
Label = paste(m1_concentration, m2_concentration, sep="x")
})
simul_cd86_p = with(filter(normalized_simultaneous_medians, antibody == "exp"), {
pairwise.t.test(CD86, Label, p.adjust.method="none")
})
(normalized_simultaneous_medians_mean %>% filter(antibody == "exp") %>%
ggplot(aes(m1_concentration, mean.CD206, group=m2_concentration)) +
facet_grid(m2_concentration~.) +
geom_point(size=0.6) +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) +
geom_path(alpha=0.3) +
ylim(0, NA) +
# ggtitle("CD206 reponse to IL-4/IL-13") +
labs(x="[LPS/IFN-γ] (ng/ml)", y="Mean CD206 intensity relative to M2") +
manuscript_theme
) %>%
print()
simul_cd206_p = with(filter(normalized_simultaneous_medians, antibody == "exp"), {
pairwise.t.test(CD206, Label, p.adjust.method="none")
})
Figure 4
Figure 4b
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(mean.CD86, mean.CD206, group=M1xM2, shape=timepoint)) +
facet_wrap(~M1xM2) +
geom_point(aes(color=timepoint), size=4) +
geom_path() +
scale_shape_manual("Timepoint", values=c(16, 17, 15, 18)) +
scale_color_discrete("Timepoint") +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), alpha=0.4) +
geom_errorbarh(aes(xmin=ymin.CD86, xmax=ymax.CD86), alpha=0.4) +
xlim(0, NA) + ylim(0, NA) +
labs(
x="CD86 intensity relative to M1",
y="CD206 intensity relative to M2"
) +
manuscript_theme
print(g)
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(log2(mean.CD86), log2(mean.CD206), group=M1xM2, shape=timepoint)) +
facet_wrap(~M1xM2) +
geom_point(aes(color=timepoint), size=4) +
geom_path() +
scale_shape_manual("Timepoint", values=c(16, 17, 15, 18)) +
scale_color_discrete("Timepoint") +
geom_errorbar(aes(ymin=log2(ymin.CD206), ymax=log2(ymax.CD206)), alpha=0.4) +
geom_errorbarh(aes(xmin=log2(ymin.CD86), xmax=log2(ymax.CD86)), alpha=0.4) +
coord_cartesian(ylim=c(-2, 2), xlim=c(-3.5, 3.5)) +
geom_hline(yintercept=0) + geom_vline(xintercept=0) +
#xlim(0, NA) + ylim(0, NA) +
theme_bw() +
theme(text=element_text(size=18)) +
labs(
x="CD86 intensity relative to M1",
y="CD206 intensity relative to M2"
)
print(g)
Figure 4c
p10b_compare = function(df) {
t24 = filter(normalized_simul_timecourse_medians,
antibody == "exp",
timepoint == "24h",
m1_concentration == df$m1_concentration[1],
m2_concentration == df$m2_concentration[1],
Experiment %in% df$Experiment)
expect_equal(t24$Experiment, df$Experiment)
t.test(t24$CD86, df$CD86, paired=TRUE)$p.value
}
p10b_p = normalized_simul_timecourse_medians %>%
filter(antibody == "exp") %>%
group_by(m1_concentration, m2_concentration, timepoint) %>%
do(p.value=p10b_compare(.)) %>%
mutate(p.value=p.value[[1]])
print(p10b_p)
# t.test(filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="24h", m1_concentration == .3, m2_concentration == 1)$CD86, mu = 1)
# t = 5.2536, df = 5, p-value = 0.003316
p10b_p = merge(p10b_p, normalized_simul_timecourse_medians_mean) %>%
filter(!is.na(p.value), p.value < 0.05, antibody == "exp")
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(timepoint, mean.CD86, color=M1xM2, group=M1xM2)) +
geom_point() +
geom_line() +
geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) +
ylim(0, NA) +
labs(x="Timepoint", y="CD86 intensity relative to M1") +
scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") +
manuscript_theme +
annotate(geom="text", label="*", size=6,
x=as.numeric(p10b_p$timepoint)+0.1,
y=p10b_p$mean.CD86)
print(g)
modeldata = readr::read_csv("../data-raw/modeling/MISA_IFFL_time_course.csv") %>%
select(timepoint=Time,
m1_concentration=`M1 inducer`,
m2_concentration=`M2 inducer`,
mean.CD86=`M1 response`,
mean.CD206=`M2 response`) %>%
mutate(M1xM2=paste(m1_concentration, m2_concentration, sep="x"),
timepoint=sprintf("%dh", timepoint))
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(timepoint, mean.CD86, color=M1xM2, group=M1xM2)) +
geom_point() +
geom_line(data=modeldata, linetype=2) +
geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) +
ylim(0, NA) +
labs(x="Timepoint", y="CD86 intensity relative to M1") +
scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") +
manuscript_theme
print(g)
Figure 4d
p10c_compare = function(df) {
t24 = filter(normalized_simul_timecourse_medians,
antibody == "exp",
timepoint == "24h",
m1_concentration == df$m1_concentration[1],
m2_concentration == df$m2_concentration[1],
Experiment %in% df$Experiment)
expect_equal(t24$Experiment, df$Experiment)
t.test(t24$CD206, df$CD206, paired=TRUE)$p.value
}
p10b_c = normalized_simul_timecourse_medians %>%
filter(antibody == "exp") %>%
group_by(m1_concentration, m2_concentration, timepoint) %>%
do(p.value=p10c_compare(.)) %>%
mutate(p.value=p.value[[1]])
print(p10b_c)
p10c_p = merge(p10b_c, normalized_simul_timecourse_medians_mean) %>%
filter(!is.na(p.value), p.value < 0.05, antibody == "exp")
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(timepoint, mean.CD206, color=M1xM2, group=M1xM2)) +
geom_point() +
geom_line() +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) +
ylim(0, NA) +
labs(x="Timepoint", y="CD206 intensity relative to M2") +
scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") +
manuscript_theme +
annotate(geom="text", label="*", size=6,
x=as.numeric(p10c_p$timepoint)+0.1,
y=p10c_p$mean.CD206)
print(g)
#t.test(filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="96h", m1_concentration == .3, m2_concentration == 1)$CD206, filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="96h", m1_concentration == 0, m2_concentration == 1)$CD206)
Is the (96h, CD206, 0x1) case different from (,,0.3x1)?
cd206_96h_m2 = (normalized_simul_timecourse_medians %>%
filter(antibody == "exp",
timepoint == "96h",
m1_concentration == 0,
m2_concentration == 1))$CD206
cd206_96h_costim = (normalized_simul_timecourse_medians %>%
filter(antibody == "exp",
timepoint == "96h",
m1_concentration == 0.3,
m2_concentration == 1))$CD206
t.test(cd206_96h_m2, cd206_96h_costim)
g = normalized_simul_timecourse_medians_mean %>%
filter(antibody == "exp") %>%
ggplot(aes(timepoint, mean.CD206, color=M1xM2, group=M1xM2)) +
geom_point() +
geom_line(data=modeldata, linetype=2) +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) +
ylim(0, NA) +
labs(x="Timepoint", y="CD206 intensity relative to M2") +
scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") +
manuscript_theme
print(g)
Figure 6
Figure 6c
# prepare the list of significant data based on the linear model below
p4sig = normalized_m1_sus_m2_medians_mean %>%
filter(m2_concentration != 0, antibody == "exp", m1_concentration %in% c(0, 0.3))
p4right = (normalized_m1_sus_m2_medians %>%
filter(antibody=="exp", m2_concentration==1, m1_concentration==0.3))
p4righttest = p4right$CD206 %>% t.test(mu=1) %>% print
p4rightlen = p4right %>% nrow %>% print
g = normalized_m1_sus_m2_medians_mean %>%
filter(antibody == "exp", m1_concentration %in% c(0, 0.3)) %>%
ggplot(aes(m2_concentration, mean.CD206, color=m1_concentration, shape=m1_concentration)) +
geom_point(size=2) +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.2) +
scale_color_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") +
scale_shape_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") +
labs(x="[IL-4/IL-13] (ng/ml), added t=24 h", y="CD206 intensity relative to M2") +
ylim(0, 1.5) +
manuscript_theme +
annotate(geom="text", label="*", size=5,
x=as.numeric(p4sig$m2_concentration)+0.15,
y=p4sig$mean.CD206) +
annotate(geom="text", label="†", size=4, x=4.4, y=1.15) +
annotate(geom="segment", x=4.3, xend=4.3, y=1, yend=1.3)
print(g)
Figure 6b
g = normalized_m1_sus_m2_medians_mean %>%
filter(antibody == "exp", m1_concentration %in% c(0, 0.3)) %>%
ggplot(aes(m2_concentration, mean.CD86, color=m1_concentration, shape=m1_concentration)) +
geom_point(size=2) +
geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.2) +
scale_color_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") +
scale_shape_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") +
labs(x="[IL-4/IL-13] (ng/ml), added t=24 h", y="CD86 intensity relative to M2") +
ylim(0, 1.6) +
manuscript_theme
print(g)
Asterisk: significantly different (p < 0.05) from no treatment
Dagger: significantly different from each other
Does M2 signal still increase from baseline in the presence of M1 stimulus? (Yes)
test4_df = normalized_m1_sus_m2_medians %>%
filter(antibody == "exp",
m1_concentration %in% c(0, 0.3))
test4_lm = lm(CD206~m1_concentration + m2_concentration:m1_concentration, test4_df)
print(summary(test4_lm))
Is M2 signal affected by the presence of M1 stimulus? (Not robustly)
test4b_results = numeric()
for(q in levels(test4_df$m2_concentration)) {
no_m1 = test4_df[test4_df$m2_concentration == q & test4_df$m1_concentration == 0,]
with_m1 = test4_df[test4_df$m2_concentration == q & test4_df$m1_concentration == 0.3,]
expect_equal(no_m1$Experiment, with_m1$Experiment) # for paired test
p = t.test(no_m1$CD206, with_m1$CD206, paired=TRUE)$p.value
test4b_results[q] = p
}
Non-adjusted:
print(test4b_results)
Adjusted (Holm, FDR):
print(p.adjust(test4b_results))
print(p.adjust(test4b_results, "fdr"))
Figure 6e
# extract points which are significant from no treatment based on the linear model below
p5sig = normalized_m2_sus_m1_medians_mean %>%
filter(antibody == "exp",
(m2_concentration == 0 & m1_concentration == 0.1) |
(m2_concentration == 0 & m1_concentration == 0.3) |
(m2_concentration == 1 & m1_concentration == 0.3))
g = normalized_m2_sus_m1_medians_mean %>%
filter(antibody == "exp", m2_concentration %in% c(0, 1)) %>%
ggplot(aes(m1_concentration, mean.CD86, color=m2_concentration, shape=m2_concentration)) +
geom_point(size=2) +
geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.2) +
scale_color_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") +
scale_shape_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") +
ylim(0, 1.6) +
labs(x="[LPS/IFN-γ] (ng/ml), added t=24 h", y="CD86 intensity relative to M1") +
manuscript_theme +
annotate(geom="text", label="*", size=5,
x=as.numeric(p5sig$m1_concentration)+0.15,
y=p5sig$mean.CD86) +
annotate(geom="text", label="†", size=4, x=2.3, y=0.3125) +
annotate(geom="segment", x=2.2, xend=2.2, y=0.149-0.04, yend=0.456+0.06)
print(g)
Figure 6f
g = normalized_m2_sus_m1_medians_mean %>%
filter(antibody == "exp", m2_concentration %in% c(0, 1)) %>%
ggplot(aes(m1_concentration, mean.CD206, color=m2_concentration, shape=m2_concentration)) +
geom_point(size=2) +
geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.2) +
scale_color_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") +
scale_shape_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") +
ylim(0, 1.5) +
labs(x="[LPS/IFN-γ] (ng/ml), added t=24 h", y="CD206 intensity relative to M2") +
manuscript_theme
print(g)
Asterisk: significantly different (p < 0.05) from no treatment
Dagger: significantly different from each other
Does M1 signal still increase from baseline in the presence of M2 stimulus? (Yes, at the high end)
test5_df = normalized_m2_sus_m1_medians %>%
filter(antibody == "exp",
m2_concentration %in% c(0, 1))
test5_lm = lm(CD86~m2_concentration + m1_concentration:m2_concentration, test5_df)
print(summary(test5_lm))
Is M1 signal affected by the presence of M2 stimulus? (Not robustly except at 0.03 ng/ml)
test5b_results = numeric()
for(q in levels(test5_df$m1_concentration)) {
no_m2 = test5_df[test5_df$m1_concentration == q & test5_df$m2_concentration == 0,]
with_m2 = test5_df[test5_df$m1_concentration == q & test5_df$m2_concentration == 1,]
expect_equal(no_m2$Experiment, with_m2$Experiment) # for paired test
p = t.test(no_m2$CD86, with_m2$CD86, paired=TRUE)$p.value
test5b_results[q] = p
}
print(p.adjust(test5b_results))
WendyLiuLab/BiactivationFlow documentation built on May 9, 2019, 10:56 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
library(dplyr) library(ggplot2) library(reshape2) library(testthat) library(FlowAnalysis) protocols = c("simultaneous", "m1_sus_m2", "m2_sus_m1") # m1_m2, m2_m1 to_save = NULL for(stub in protocols) { protocol = eval(parse(text=stub)) normalized = normalize_by_positive_controls(protocol) normalized_medians = as.data.frame(make_medians(normalized)) normalized_medians_mean = normalized_medians %>% group_by(antibody, m1_concentration, m2_concentration) %>% summarize(mean.CD86=mean(CD86), sd.CD86=sd(CD86), mean.CD206=mean(CD206), sd.CD206=sd(CD206), n=n()) %>% mutate(sem.CD86=sd.CD86/sqrt(n), sem.CD206=sd.CD206/sqrt(n)) %>% mutate(ymax.CD86=mean.CD86+sem.CD86, ymin.CD86=mean.CD86-sem.CD86, ymax.CD206=mean.CD206+sem.CD206, ymin.CD206=mean.CD206-sem.CD206) eval(parse(text=sprintf("normalized_%s = normalized", stub))) eval(parse(text=sprintf("normalized_%s_medians = normalized_medians", stub))) eval(parse(text=sprintf("normalized_%s_medians_mean = normalized_medians_mean", stub))) to_save = c(to_save, # sprintf("normalized_%s", stub), sprintf("normalized_%s_medians", stub), sprintf("normalized_%s_medians_mean", stub)) } # simul_timecourse is special normalized_simul_timecourse_medians = simul_timecourse %>% normalize_by_positive_controls_timepoint %>% group_by(Experiment, antibody, timepoint, m1_concentration, m2_concentration) %>% summarize(CD206 = median(CD206), CD86 = median(CD86)) normalized_simul_timecourse_medians_mean = normalized_simul_timecourse_medians %>% group_by(m1_concentration, m2_concentration, timepoint, antibody) %>% summarize(mean.CD86=mean(CD86), sd.CD86=sd(CD86), mean.CD206=mean(CD206), sd.CD206=sd(CD206), n=n()) %>% mutate(sem.CD86=sd.CD86/sqrt(n), sem.CD206=sd.CD206/sqrt(n)) %>% mutate(ymax.CD86=mean.CD86+sem.CD86, ymin.CD86=mean.CD86-sem.CD86, ymax.CD206=mean.CD206+sem.CD206, ymin.CD206=mean.CD206-sem.CD206, M1xM2=paste(m1_concentration, m2_concentration, sep="x")) manuscript_theme = theme_bw() + theme(text=element_text(size=8))
Figure 1
1a. M1 dose-response histogram
g = simultaneous %>% filter(Experiment == "tds-004-03", m2_concentration == 0) %>% mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>% ggplot(aes(CD86, color=antibody)) + facet_grid(m1_concentration~.) + geom_density() + scale_color_discrete("Antibody") + scale_x_continuous(trans=biexp_trans(lim=100, decade.size=400), breaks=c(-100, 0,100,1000,10000,100000), labels=c("-100", "0","100","1000","1e4","1e5")) + scale_y_continuous(limits=c(0, 0.006), breaks=c(0, 0.003, 0.006)) + labs(x="CD86 intensity", y="Density") + geom_vline( data=(simultaneous %>% filter(Experiment == "tds-004-03", m2_concentration == 0) %>% mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>% group_by(Experiment, m1_concentration, antibody) %>% summarize(median.CD86=median(CD86))), mapping=aes(xintercept=median.CD86, color=antibody), alpha=0.5 ) + guides(color=guide_legend(direction="horizontal")) + manuscript_theme + theme(legend.position="bottom") print(g)
1b. M1 response increases with M1 stimulus
plot_1_filter = function(x) filter(x, antibody == "exp", m2_concentration == 0) g = normalized_simultaneous_medians %>% plot_1_filter %>% ggplot(aes(m1_concentration)) + geom_point(aes(y=CD86), alpha=0.2) + geom_line(aes(y=CD86, group=Experiment), alpha=0.2) + geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean), aes(ymax=ymax.CD86, ymin=ymin.CD86), width=0.2) + geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean), aes(ymax=mean.CD86, ymin=mean.CD86), width=0.15) + scale_x_discrete("[LPS/IFN-γ] (ng/ml)") + scale_y_continuous("CD86 intensity relative to M1", limits=c(0, 1.1), breaks=seq(0, 1, 0.2)) + manuscript_theme + annotate(geom="text", x=c(3,4)+.15, y=c(.67,1.01), label="*", size=6) print(g)
plot_1_filter = function(x) filter(x, antibody == "exp", m2_concentration == 0) g = normalized_simultaneous_medians %>% plot_1_filter %>% ggplot(aes(m1_concentration)) + geom_point(aes(y=CD86), alpha=0.2) + geom_line(aes(y=CD86, group=Experiment), alpha=0.2) + geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean), aes(ymax=ymax.CD86, ymin=ymin.CD86), width=0.2) + geom_errorbar(data=plot_1_filter(normalized_simultaneous_medians_mean), aes(ymax=mean.CD86, ymin=mean.CD86), width=0.15) + scale_x_discrete("[LPS/IFN-γ] (ng/ml)") + scale_y_continuous("CD86 intensity relative to M1", limits=c(0, 1.1), breaks=seq(0, 1, 0.2)) + manuscript_theme + annotate(geom="text", x=c(3,4)+.15, y=c(.67,1.01), label="*", size=6) print(g)
Post-hoc pairwise significance:
p1f = plot_1_filter(normalized_simultaneous_medians) plot_1_anova = aov(CD86~m1_concentration, p1f) print(summary(plot_1_anova)) print(pairwise.t.test(p1f$CD86, p1f$m1_concentration, "holm"))
Linear regression with dummy coding (significance means different from no treatment):
print(summary(lm(CD86~m1_concentration, p1f)))
1c. M2 dose-response histogram
g = simultaneous %>% filter(Experiment == "tds-004-03", m1_concentration == 0) %>% mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>% ggplot(aes(CD206, color=antibody)) + facet_grid(m2_concentration~.) + geom_density() + scale_color_discrete("Antibody") + scale_x_continuous(trans=biexp_trans(lim=100, decade.size=400), breaks=c(-100, 0,100,1000,10000,100000), labels=c("-100", "0","100","1000","1e4","1e5"), limits=c(-100,NA)) + scale_y_continuous(limits=c(0, 0.006), breaks=c(0, 0.003, 0.006)) + labs(x="CD206 intensity", y="Density") + geom_vline( data=(simultaneous %>% filter(Experiment == "tds-004-03", m1_concentration == 0) %>% mutate(antibody=factor(antibody, levels=c("exp", "iso", "unstained"), labels=c("Experimental", "Isotype", "Unstained"))) %>% group_by(Experiment, m2_concentration, antibody) %>% summarize(median.CD206=median(CD206))), mapping=aes(xintercept=median.CD206, color=antibody), alpha=0.5 ) + guides(color=guide_legend(direction="horizontal")) + manuscript_theme + theme(legend.position="bottom") print(g)
How much does M1 response increase over baseline?
f1_baseline = normalized_simultaneous_medians %>% filter(antibody == "exp", m1_concentration==0, m2_concentration==0) f1_cd86_t = t.test(f1_baseline$CD86, mu=1) print(1/f1_cd86_t$estimate) print(1/f1_cd86_t$conf.int)
1d. M2 response increases with M2 stimulus
plot_2_filter = function(x) filter(x, antibody == "exp", m1_concentration == 0) g = normalized_simultaneous_medians %>% plot_2_filter %>% ggplot(aes(m2_concentration)) + geom_point(aes(y=CD206), alpha=0.2) + geom_line(aes(y=CD206, group=Experiment), alpha=0.2) + geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean), aes(ymax=ymax.CD206, ymin=ymin.CD206), width=0.2) + geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean), aes(ymax=mean.CD206, ymin=mean.CD206), width=0.15) + scale_x_discrete("[IL-4/IL-13] (ng/ml)") + scale_y_continuous("CD206 intensity relative to M2", limits=c(0, 1.1), breaks=seq(0, 1, 0.2)) + manuscript_theme + annotate(geom="text", x=c(3.05, 4)+.12, y=c(.59, 1.01), label="*", size=6) print(g)
plot_2_filter = function(x) filter(x, antibody == "exp", m1_concentration == 0) g = normalized_simultaneous_medians %>% plot_2_filter %>% ggplot(aes(m2_concentration)) + geom_point(aes(y=CD206), alpha=0.2) + geom_line(aes(y=CD206, group=Experiment), alpha=0.2) + geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean), aes(ymax=ymax.CD206, ymin=ymin.CD206), width=0.2) + geom_errorbar(data=plot_2_filter(normalized_simultaneous_medians_mean), aes(ymax=mean.CD206, ymin=mean.CD206), width=0.15) + scale_x_discrete("[IL-4/IL-13] (ng/ml)") + scale_y_continuous("CD206 intensity relative to M2", limits=c(0, 1.1), breaks=seq(0, 1, 0.2)) + manuscript_theme + annotate(geom="text", x=c(3.05, 4)+.12, y=c(.59, 1.01), label="*", size=6) print(g)
Post-hoc pairwise significance:
p2f = plot_2_filter(normalized_simultaneous_medians) plot_2_anova = aov(CD206~m2_concentration, p2f) print(summary(plot_2_anova)) print(pairwise.t.test(p2f$CD206, p2f$m2_concentration, "holm"))
Linear regression with dummy coding (significance means different from no treatment):
print(summary(lm(CD206~m2_concentration, p2f)))
How much does M2 response increase over baseline?
f1_cd206_t = t.test(f1_baseline$CD206, mu=1) print(1/f1_cd206_t$estimate) print(1/f1_cd206_t$conf.int)
Figure 2
Figure 2b
g = normalized_simultaneous %>% filter(antibody == "exp", Experiment == "tds-003-94") %>% ggplot(aes(log2(CD86), log2(CD206))) + stat_density2d(aes(alpha=..level..), geom="polygon", n=256) + # xlim(-1, NA) + # ylim(-1, NA) + # scale_y_continuous(breaks=c(0, 2.5, 5), limits=c(-1, NA)) + xlim(-10,5) + ylim(-4,6) + geom_hline(yintercept=0, alpha=0.5) + geom_vline(xintercept=0, alpha=0.5) + coord_cartesian(xlim=c(-6,6), ylim=c(-4,4)) + facet_grid(m2_concentration~m1_concentration) + scale_alpha_continuous("Density", trans="sqrt", guide="none") + labs(x="CD86 intensity relative to M1", y="CD206 intensity relative to M2") + manuscript_theme + geom_point(data=filter(normalized_simultaneous_medians, Experiment=="tds-003-94", antibody=="exp"), mapping=aes(log2(CD86), log2(CD206)), color="red", size=0.6) print(g)
Figure 2c, 2d
p14.cd86.ann = data.frame( m1_concentration=c("0.1", "0.3", "0.3"), x=c(1, 1, 2), xend=c(4, 2, 4), y=c(2, 1.9, 2.2) ) (normalized_simultaneous_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(m2_concentration, mean.CD86, group=m1_concentration)) + facet_grid(m1_concentration~.) + geom_point(size=0.6) + geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) + geom_path(alpha=0.3) + ylim(0, 2.5) + # ggtitle("CD86 vs M2 dose, by M1 dose") + labs(x="[IL-4/IL-13] (ng/ml)", y="Mean CD86 intensity relative to M1") + manuscript_theme + geom_segment(data=p14.cd86.ann, mapping=aes(x=x, xend=xend, y=y, yend=y), show.legend=FALSE) + geom_text(data=p14.cd86.ann, mapping=aes(x=(x+xend)/2, y=y+0.08), label="*", show.legend=FALSE, size=4) + geom_segment(data=p14.cd86.ann, mapping=aes(x=x, xend=x, y=y, yend=y-0.1), show.legend=FALSE) + geom_segment(data=p14.cd86.ann, mapping=aes(x=xend, xend=xend, y=y, yend=y-0.1), show.legend=FALSE) ) %>% print normalized_simultaneous_medians = within(normalized_simultaneous_medians, { Label = paste(m1_concentration, m2_concentration, sep="x") }) simul_cd86_p = with(filter(normalized_simultaneous_medians, antibody == "exp"), { pairwise.t.test(CD86, Label, p.adjust.method="none") }) (normalized_simultaneous_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(m1_concentration, mean.CD206, group=m2_concentration)) + facet_grid(m2_concentration~.) + geom_point(size=0.6) + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) + geom_path(alpha=0.3) + ylim(0, NA) + # ggtitle("CD206 reponse to IL-4/IL-13") + labs(x="[LPS/IFN-γ] (ng/ml)", y="Mean CD206 intensity relative to M2") + manuscript_theme ) %>% print() simul_cd206_p = with(filter(normalized_simultaneous_medians, antibody == "exp"), { pairwise.t.test(CD206, Label, p.adjust.method="none") })
Figure 4
Figure 4b
g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(mean.CD86, mean.CD206, group=M1xM2, shape=timepoint)) + facet_wrap(~M1xM2) + geom_point(aes(color=timepoint), size=4) + geom_path() + scale_shape_manual("Timepoint", values=c(16, 17, 15, 18)) + scale_color_discrete("Timepoint") + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), alpha=0.4) + geom_errorbarh(aes(xmin=ymin.CD86, xmax=ymax.CD86), alpha=0.4) + xlim(0, NA) + ylim(0, NA) + labs( x="CD86 intensity relative to M1", y="CD206 intensity relative to M2" ) + manuscript_theme print(g) g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(log2(mean.CD86), log2(mean.CD206), group=M1xM2, shape=timepoint)) + facet_wrap(~M1xM2) + geom_point(aes(color=timepoint), size=4) + geom_path() + scale_shape_manual("Timepoint", values=c(16, 17, 15, 18)) + scale_color_discrete("Timepoint") + geom_errorbar(aes(ymin=log2(ymin.CD206), ymax=log2(ymax.CD206)), alpha=0.4) + geom_errorbarh(aes(xmin=log2(ymin.CD86), xmax=log2(ymax.CD86)), alpha=0.4) + coord_cartesian(ylim=c(-2, 2), xlim=c(-3.5, 3.5)) + geom_hline(yintercept=0) + geom_vline(xintercept=0) + #xlim(0, NA) + ylim(0, NA) + theme_bw() + theme(text=element_text(size=18)) + labs( x="CD86 intensity relative to M1", y="CD206 intensity relative to M2" ) print(g)
Figure 4c
p10b_compare = function(df) { t24 = filter(normalized_simul_timecourse_medians, antibody == "exp", timepoint == "24h", m1_concentration == df$m1_concentration[1], m2_concentration == df$m2_concentration[1], Experiment %in% df$Experiment) expect_equal(t24$Experiment, df$Experiment) t.test(t24$CD86, df$CD86, paired=TRUE)$p.value } p10b_p = normalized_simul_timecourse_medians %>% filter(antibody == "exp") %>% group_by(m1_concentration, m2_concentration, timepoint) %>% do(p.value=p10b_compare(.)) %>% mutate(p.value=p.value[[1]]) print(p10b_p) # t.test(filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="24h", m1_concentration == .3, m2_concentration == 1)$CD86, mu = 1) # t = 5.2536, df = 5, p-value = 0.003316 p10b_p = merge(p10b_p, normalized_simul_timecourse_medians_mean) %>% filter(!is.na(p.value), p.value < 0.05, antibody == "exp") g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(timepoint, mean.CD86, color=M1xM2, group=M1xM2)) + geom_point() + geom_line() + geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) + ylim(0, NA) + labs(x="Timepoint", y="CD86 intensity relative to M1") + scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") + manuscript_theme + annotate(geom="text", label="*", size=6, x=as.numeric(p10b_p$timepoint)+0.1, y=p10b_p$mean.CD86) print(g)
modeldata = readr::read_csv("../data-raw/modeling/MISA_IFFL_time_course.csv") %>% select(timepoint=Time, m1_concentration=`M1 inducer`, m2_concentration=`M2 inducer`, mean.CD86=`M1 response`, mean.CD206=`M2 response`) %>% mutate(M1xM2=paste(m1_concentration, m2_concentration, sep="x"), timepoint=sprintf("%dh", timepoint)) g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(timepoint, mean.CD86, color=M1xM2, group=M1xM2)) + geom_point() + geom_line(data=modeldata, linetype=2) + geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.15) + ylim(0, NA) + labs(x="Timepoint", y="CD86 intensity relative to M1") + scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") + manuscript_theme print(g)
Figure 4d
p10c_compare = function(df) { t24 = filter(normalized_simul_timecourse_medians, antibody == "exp", timepoint == "24h", m1_concentration == df$m1_concentration[1], m2_concentration == df$m2_concentration[1], Experiment %in% df$Experiment) expect_equal(t24$Experiment, df$Experiment) t.test(t24$CD206, df$CD206, paired=TRUE)$p.value } p10b_c = normalized_simul_timecourse_medians %>% filter(antibody == "exp") %>% group_by(m1_concentration, m2_concentration, timepoint) %>% do(p.value=p10c_compare(.)) %>% mutate(p.value=p.value[[1]]) print(p10b_c) p10c_p = merge(p10b_c, normalized_simul_timecourse_medians_mean) %>% filter(!is.na(p.value), p.value < 0.05, antibody == "exp") g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(timepoint, mean.CD206, color=M1xM2, group=M1xM2)) + geom_point() + geom_line() + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) + ylim(0, NA) + labs(x="Timepoint", y="CD206 intensity relative to M2") + scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") + manuscript_theme + annotate(geom="text", label="*", size=6, x=as.numeric(p10c_p$timepoint)+0.1, y=p10c_p$mean.CD206) print(g) #t.test(filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="96h", m1_concentration == .3, m2_concentration == 1)$CD206, filter(normalized_simul_timecourse_medians, antibody=="exp", timepoint=="96h", m1_concentration == 0, m2_concentration == 1)$CD206)
Is the (96h, CD206, 0x1) case different from (,,0.3x1)?
cd206_96h_m2 = (normalized_simul_timecourse_medians %>% filter(antibody == "exp", timepoint == "96h", m1_concentration == 0, m2_concentration == 1))$CD206 cd206_96h_costim = (normalized_simul_timecourse_medians %>% filter(antibody == "exp", timepoint == "96h", m1_concentration == 0.3, m2_concentration == 1))$CD206 t.test(cd206_96h_m2, cd206_96h_costim)
g = normalized_simul_timecourse_medians_mean %>% filter(antibody == "exp") %>% ggplot(aes(timepoint, mean.CD206, color=M1xM2, group=M1xM2)) + geom_point() + geom_line(data=modeldata, linetype=2) + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.15) + ylim(0, NA) + labs(x="Timepoint", y="CD206 intensity relative to M2") + scale_color_discrete("[LPS/IFN-γ] x\n[IL-4/IL-13]\n(ng/ml)") + manuscript_theme print(g)
Figure 6
Figure 6c
# prepare the list of significant data based on the linear model below p4sig = normalized_m1_sus_m2_medians_mean %>% filter(m2_concentration != 0, antibody == "exp", m1_concentration %in% c(0, 0.3)) p4right = (normalized_m1_sus_m2_medians %>% filter(antibody=="exp", m2_concentration==1, m1_concentration==0.3)) p4righttest = p4right$CD206 %>% t.test(mu=1) %>% print p4rightlen = p4right %>% nrow %>% print g = normalized_m1_sus_m2_medians_mean %>% filter(antibody == "exp", m1_concentration %in% c(0, 0.3)) %>% ggplot(aes(m2_concentration, mean.CD206, color=m1_concentration, shape=m1_concentration)) + geom_point(size=2) + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.2) + scale_color_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") + scale_shape_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") + labs(x="[IL-4/IL-13] (ng/ml), added t=24 h", y="CD206 intensity relative to M2") + ylim(0, 1.5) + manuscript_theme + annotate(geom="text", label="*", size=5, x=as.numeric(p4sig$m2_concentration)+0.15, y=p4sig$mean.CD206) + annotate(geom="text", label="†", size=4, x=4.4, y=1.15) + annotate(geom="segment", x=4.3, xend=4.3, y=1, yend=1.3) print(g)
Figure 6b
g = normalized_m1_sus_m2_medians_mean %>% filter(antibody == "exp", m1_concentration %in% c(0, 0.3)) %>% ggplot(aes(m2_concentration, mean.CD86, color=m1_concentration, shape=m1_concentration)) + geom_point(size=2) + geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.2) + scale_color_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") + scale_shape_discrete("[LPS/IFN-γ]\n(ng/ml)\nadded t=0 h") + labs(x="[IL-4/IL-13] (ng/ml), added t=24 h", y="CD86 intensity relative to M2") + ylim(0, 1.6) + manuscript_theme print(g)
Asterisk: significantly different (p < 0.05) from no treatment Dagger: significantly different from each other
Does M2 signal still increase from baseline in the presence of M1 stimulus? (Yes)
test4_df = normalized_m1_sus_m2_medians %>% filter(antibody == "exp", m1_concentration %in% c(0, 0.3)) test4_lm = lm(CD206~m1_concentration + m2_concentration:m1_concentration, test4_df) print(summary(test4_lm))
Is M2 signal affected by the presence of M1 stimulus? (Not robustly)
test4b_results = numeric() for(q in levels(test4_df$m2_concentration)) { no_m1 = test4_df[test4_df$m2_concentration == q & test4_df$m1_concentration == 0,] with_m1 = test4_df[test4_df$m2_concentration == q & test4_df$m1_concentration == 0.3,] expect_equal(no_m1$Experiment, with_m1$Experiment) # for paired test p = t.test(no_m1$CD206, with_m1$CD206, paired=TRUE)$p.value test4b_results[q] = p }
Non-adjusted:
print(test4b_results)
Adjusted (Holm, FDR):
print(p.adjust(test4b_results)) print(p.adjust(test4b_results, "fdr"))
Figure 6e
# extract points which are significant from no treatment based on the linear model below p5sig = normalized_m2_sus_m1_medians_mean %>% filter(antibody == "exp", (m2_concentration == 0 & m1_concentration == 0.1) | (m2_concentration == 0 & m1_concentration == 0.3) | (m2_concentration == 1 & m1_concentration == 0.3)) g = normalized_m2_sus_m1_medians_mean %>% filter(antibody == "exp", m2_concentration %in% c(0, 1)) %>% ggplot(aes(m1_concentration, mean.CD86, color=m2_concentration, shape=m2_concentration)) + geom_point(size=2) + geom_errorbar(aes(ymin=ymin.CD86, ymax=ymax.CD86), width=0.2) + scale_color_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") + scale_shape_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") + ylim(0, 1.6) + labs(x="[LPS/IFN-γ] (ng/ml), added t=24 h", y="CD86 intensity relative to M1") + manuscript_theme + annotate(geom="text", label="*", size=5, x=as.numeric(p5sig$m1_concentration)+0.15, y=p5sig$mean.CD86) + annotate(geom="text", label="†", size=4, x=2.3, y=0.3125) + annotate(geom="segment", x=2.2, xend=2.2, y=0.149-0.04, yend=0.456+0.06) print(g)
Figure 6f
g = normalized_m2_sus_m1_medians_mean %>% filter(antibody == "exp", m2_concentration %in% c(0, 1)) %>% ggplot(aes(m1_concentration, mean.CD206, color=m2_concentration, shape=m2_concentration)) + geom_point(size=2) + geom_errorbar(aes(ymin=ymin.CD206, ymax=ymax.CD206), width=0.2) + scale_color_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") + scale_shape_discrete("[IL-4/IL-13]\n(ng/ml)\nadded t=0 h") + ylim(0, 1.5) + labs(x="[LPS/IFN-γ] (ng/ml), added t=24 h", y="CD206 intensity relative to M2") + manuscript_theme print(g)
Asterisk: significantly different (p < 0.05) from no treatment Dagger: significantly different from each other
Does M1 signal still increase from baseline in the presence of M2 stimulus? (Yes, at the high end)
test5_df = normalized_m2_sus_m1_medians %>% filter(antibody == "exp", m2_concentration %in% c(0, 1)) test5_lm = lm(CD86~m2_concentration + m1_concentration:m2_concentration, test5_df) print(summary(test5_lm))
Is M1 signal affected by the presence of M2 stimulus? (Not robustly except at 0.03 ng/ml)
test5b_results = numeric() for(q in levels(test5_df$m1_concentration)) { no_m2 = test5_df[test5_df$m1_concentration == q & test5_df$m2_concentration == 0,] with_m2 = test5_df[test5_df$m1_concentration == q & test5_df$m2_concentration == 1,] expect_equal(no_m2$Experiment, with_m2$Experiment) # for paired test p = t.test(no_m2$CD86, with_m2$CD86, paired=TRUE)$p.value test5b_results[q] = p } print(p.adjust(test5b_results))
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.