Figure 5: Comparison of H170 variants

Figure 5 uses panel B to display maximum amplitude and decay rate constants for each of the H170 mutants at LPR5 and LPR10.

The Dataset

We process the NeRh50_Table_S1_raw_data_used_in_figures.csv dataset to obtain the data for this analysis in Table 1.

# Show interactive tables of data
DT::datatable(dfms$dfm_fig5_amp)
DT::datatable(dfms$dfm_fig5_k)
# Write datasets to file (commented out for website)
#readr::write_csv(dfms$dfm_fig5_amp, "assets/data/03_fig5_amplitude.csv")
#readr::write_csv(dfms$dfm_fig5_k, "assets/data/03_fig5_rate.csv")

Table 1: Measurements for WT and H170 variants across LPR5 and LPR10 for maximum amplitude and decay rate constant.

(a) Maximum amplitude

(b) Decay rate

We visualise the dataset in the style of draft figure 5, in Figure 1.

# Visualise datasets as box/whisker with superimposed datapoints
p4_h170a_amp <- ggplot2::ggplot(dfms$dfm_fig5_amp |> dplyr::filter(Variant=="H170A"), 
                                ggplot2::aes(x=LPR, y=Value, color=LPR))  +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Maximum amplitude (nA)", x="") +
  ggplot2::ylim(0, 4) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))

p4_h170a_k <- ggplot2::ggplot(dfms$dfm_fig5_k |> dplyr::filter(Variant == "H170A"), 
                              ggplot2::aes(x=LPR, y=Value, color=LPR))  +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Decay rate constant (/s)", x="") +
  ggplot2::ylim(0, 75) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))

p4_h170d_amp <- ggplot2::ggplot(dfms$dfm_fig5_amp |> dplyr::filter(Variant == "H170D"),
                                ggplot2::aes(x=LPR, y=Value, color=LPR)) +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Maximum amplitude (nA)", x="") +
  ggplot2::ylim(0, 4) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))

p4_h170d_k <- ggplot2::ggplot(dfms$dfm_fig5_k |> dplyr::filter(Variant == "H170D"), 
                              ggplot2::aes(x=LPR, y=Value, color=LPR))  +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Decay rate constant (/s)", x="") +
  ggplot2::ylim(0, 75) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))

p4_h170e_amp <- ggplot2::ggplot(dfms$dfm_fig5_amp |> dplyr::filter(Variant == "H170E"), 
                                ggplot2::aes(x=LPR, y=Value, color=LPR))  +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Maximum amplitude (nA)", x="") +
  ggplot2::ylim(0, 4) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))

p4_h170e_k <- ggplot2::ggplot(dfms$dfm_fig5_k |> dplyr::filter(Variant == "H170E"),
                              ggplot2::aes(x=LPR, y=Value, color=LPR))  +
  ggplot2::geom_boxplot(outlier.shape = NA) +
  ggplot2::geom_jitter() +
  ggplot2::labs(y="Decay rate constant (/s)", x="") +
  ggplot2::ylim(0, 75) +
  ggplot2::theme_minimal() + 
  ggplot2::theme(axis.text.x = ggplot2::element_text(angle = 45, vjust = 0.5, hjust=1))
  
ggpubr::ggarrange(p4_h170a_amp, p4_h170a_k, p4_h170d_amp, p4_h170d_k, p4_h170e_amp, p4_h170e_k,
                  common.legend=TRUE,
                  labels=c("H170A", "", "H170D", "", "H170E", ""),
                  nrow=1
                  )

Figure 1: H170 variant measurements across LPR5 and LPR10 as in figure 5B from the draft manuscript

Downloads

• Download 03_fig5_amplitude.csv
• Download 03_fig5_rate.csv

The Models

We treat the system as a single experiment with two dependent variables (maximum amplitude and decay rate constant), fitting linear models with LPR and Variant as explanatory variables. We initially fit the full model Value ~ LPR * Variant as in Table 2.

model4_amp_lm <- lm(Value ~ LPR * Variant, data=dfms$dfm_fig5_amp)
model4_k_lm <- lm(Value ~ LPR * Variant, data=dfms$dfm_fig5_k)

gtsummary::tbl_regression(model4_amp_lm)
gtsummary::tbl_regression(model4_k_lm)

Table 2: Linear model fits of Value ~ LPR * Variant

(a) Maximum amplitude

Characteristic	Beta	95% CI	p-value
LPR
LPR5	—	—
LPR10	-0.91	-1.3, -0.55	<0.001
Variant
WT	—	—
H170A	-0.58	-0.96, -0.20	0.003
H170D	-0.33	-0.68, 0.02	0.066
H170E	0.33	-0.03, 0.70	0.075
LPR * Variant
LPR10 * H170A	0.45	-0.12, 1.0	0.12
LPR10 * H170D	0.10	-0.48, 0.69	0.7
LPR10 * H170E	-0.31	-0.85, 0.23	0.3
Abbreviation: CI = Confidence Interval

(b) Decay rate

Characteristic	Beta	95% CI	p-value
LPR
LPR5	—	—
LPR10	-11	-15, -6.8	<0.001
Variant
WT	—	—
H170A	-8.5	-13, -3.9	<0.001
H170D	-13	-17, -8.8	<0.001
H170E	-20	-25, -16	<0.001
LPR * Variant
LPR10 * H170A	1.3	-6.3, 8.9	0.7
LPR10 * H170D	-0.54	-7.7, 6.7	0.9
LPR10 * H170E	0.73	-5.5, 7.0	0.8
Abbreviation: CI = Confidence Interval

The model fits in Table 2 indicate that there are no important interaction effects for either maximum amplitude or decay rate constant.

The models indicate the following non-zero differences in maximum amplitude:

• between LPR5 and LPR10 (-0.91nA [-1.3, -0.55])
• between WT and H170A (-0.58nA [-0.96, -0.20])

They also indicate non-zero differences in decay rate constant:

• between LPR5 and LPR10 (-11/s [-15, -6.8])
• between WT and all three H170 variants (a reduction in decay rate constant):
  • H170A, -8.5/s [-13, -3.9]
  • H170D -13/s [-17, -8.8]
  • H170E, -20/s [-25, -16].

We check the model fits in Figure 2.

performance::check_model(model4_amp_lm, size_title=6)

performance::check_model(model4_k_lm, size_title=6)

(a) Maximum amplitude

(b) Decay rate

Figure 2: Diagnostic plots of Value ~ LPR + Variant fit

On the whole the diagnostic plots do not indicate issues, except for the LPR:Variant correlation. However, as this is an interaction term, the collinearity is likely a result of “essential ill-conditioning” and unavoidable. As the interactions are not indicated to have any great effect on the model we proceed with the model as it stands, and extract the contrasts in Table 3.

EMM1 <- emmeans::emmeans(model4_amp_lm, ~ LPR | Variant)
confint(emmeans::contrast(EMM1, "pairwise", adjust="sidak")) |>
  kableExtra::kbl(digits=3) |>
  kableExtra::kable_styling(
    bootstrap_options = c("hover", "responsive")
  )
EMM2 <- emmeans::emmeans(model4_k_lm, ~ LPR | Variant)
confint(emmeans::contrast(EMM2, "pairwise", adjust="sidak")) |>
  kableExtra::kbl(digits=3) |>
  kableExtra::kable_styling(
    bootstrap_options = c("hover", "responsive")
  )

Table 3: Contrasts for model fits, with Sidak adjustment

(a) Maximum amplitude: change `LPR`, constant `Variant`

contrast	Variant	estimate	SE	df	lower.CL	upper.CL
LPR5 - LPR10	WT	0.914	0.183	54	0.546	1.281
LPR5 - LPR10	H170A	0.459	0.221	54	0.015	0.903
LPR5 - LPR10	H170D	0.810	0.227	54	0.355	1.265
LPR5 - LPR10	H170E	1.225	0.196	54	0.832	1.617

(b) Decay rate: change `LPR`, constant `Variant`

contrast	Variant	estimate	SE	df	lower.CL	upper.CL
LPR5 - LPR10	WT	11.023	2.125	52	6.759	15.286
LPR5 - LPR10	H170A	9.725	3.110	52	3.484	15.966
LPR5 - LPR10	H170D	11.563	2.887	52	5.770	17.356
LPR5 - LPR10	H170E	10.290	2.271	52	5.733	14.848

Restricting interpretation to the contrasts in manuscript figure 5, we find the following robustly nonzero differences for LPR changes:

• Maximum amplitude decreases for H170A by 0.459nA [0.0151,0.903] from LPR5 to LPR10
• Maximum amplitude decreases for H170D by 0.810nA [0.355,1.26] from LPR5 to LPR10
• Maximum amplitude decreases for H170E by 1.22nA [0.832,1.62] from LPR5 to LPR10
• Decay rate decreases for H170A by 9.72/s [3.48,16.0] from LPR5 to LPR10
• Decay rate decreases for H170A by 11.6/s [5.77,17.4] from LPR5 to LPR10
• Decay rate decreases for H170A by 10.3 [5.73,14.8] from LPR5 to LPR10

Figure

We visualise this with \(p\)-values in Figure 3, which we interpret as the change from LPR5 to LPR10 inducing a significant reduction in maximum amplitude and decay rate constant for all three variants.

# Get p-values, accounting for heteroscedasticity and multiple testing
# Also generate group1 and group2 columns
EMM_amp <- as.data.frame(pairs(EMM1)) |>
  tidyr::separate(contrast, into=c("group1", "group2"), sep=" - ") |>
  dplyr::mutate(y.position=4) |>
  rstatix::add_significance()
EMM_k <- as.data.frame(pairs(EMM2)) |>
  tidyr::separate(contrast, into=c("group1", "group2"), sep=" - ") |>
  dplyr::mutate(y.position=70) |>
  rstatix::add_significance()

p4_h170a_amp_annot <- p4_h170a_amp + 
  ggpubr::stat_pvalue_manual(EMM_amp |> dplyr::filter(Variant == "H170A"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1) 
p4_h170a_k_annot <- p4_h170a_k + 
  ggpubr::stat_pvalue_manual(EMM_k |> dplyr::filter(Variant == "H170A"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1) 
p4_h170d_amp_annot <- p4_h170d_amp + 
  ggpubr::stat_pvalue_manual(EMM_amp |> dplyr::filter(Variant == "H170D"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1)
p4_h170d_k_annot <- p4_h170d_k + 
  ggpubr::stat_pvalue_manual(EMM_k |> dplyr::filter(Variant == "H170D"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1) 
p4_h170e_amp_annot <- p4_h170e_amp + 
  ggpubr::stat_pvalue_manual(EMM_amp |> dplyr::filter(Variant == "H170E"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1)
p4_h170e_k_annot <- p4_h170e_k + 
  ggpubr::stat_pvalue_manual(EMM_k |> dplyr::filter(Variant == "H170E"), label="p = {scales::pvalue(p.value)}",
                             size=3, step.increase=0.1)


p_comp <- ggpubr::ggarrange(p4_h170a_amp_annot, p4_h170a_k_annot,
                            p4_h170d_amp_annot, p4_h170d_k_annot,
                            p4_h170e_amp_annot, p4_h170e_k_annot,
                            common.legend=TRUE,
                            labels=c("H170A", "", "H170D", "", "H170E", ""), vjust=5,
                            legend="bottom",
                            nrow=1)
p_comp

# Write figures to file (commented out for webpage)
#ggplot2::ggsave("assets/images/fig05_H170_mutants.pdf", p_comp)
#ggplot2::ggsave("assets/images/fig05_H170_mutants.png", p_comp)

Figure 3: H170 variant measurements across LPR5 and LPR10 as in figure 5B from the draft manuscript , with Sidak-adjusted p-values indicated

Downloads

• Download as .pdf
• Download as .png

Additional Contrasts

There are further contrasts used as evidence in the manuscript text. The question arises whether there is a significant difference between WT and H170 variant maximum amplitude or decay rate constant at LPR5, which we explore in Table 4. This indicates a difference in maximum amplitude between WT and H170A (but not H170D or H170E), and a difference in decay rate constant between WT and all H170 variants, at LPR5.

EMM1 <- emmeans::emmeans(model4_amp_lm, ~ Variant | LPR)
confint(emmeans::contrast(EMM1, "pairwise", adjust="sidak")) |>
  kableExtra::kbl(digits=3) |>
  kableExtra::kable_styling(
    bootstrap_options = c("hover", "responsive")
  )
EMM2 <- emmeans::emmeans(model4_k_lm, ~ Variant | LPR)
confint(emmeans::contrast(EMM2, "pairwise", adjust="sidak")) |>
  kableExtra::kbl(digits=3) |>
  kableExtra::kable_styling(
    bootstrap_options = c("hover", "responsive")
  )

Table 4: Further contrasts for model fits

(a) Maximum amplitude: changes between `Variant` at each `LPR`

contrast	LPR	estimate	SE	df	lower.CL	upper.CL
WT - H170A	LPR5	0.580	0.189	54	0.064	1.095
WT - H170D	LPR5	0.328	0.175	54	-0.149	0.805
WT - H170E	LPR5	-0.332	0.183	54	-0.832	0.168
H170A - H170D	LPR5	-0.252	0.181	54	-0.745	0.241
H170A - H170E	LPR5	-0.912	0.189	54	-1.427	-0.396
H170D - H170E	LPR5	-0.660	0.175	54	-1.137	-0.183
WT - H170A	LPR10	0.125	0.217	54	-0.467	0.717
WT - H170D	LPR10	0.224	0.233	54	-0.413	0.862
WT - H170E	LPR10	-0.021	0.196	54	-0.556	0.513
H170A - H170D	LPR10	0.099	0.261	54	-0.612	0.811
H170A - H170E	LPR10	-0.146	0.227	54	-0.768	0.475
H170D - H170E	LPR10	-0.246	0.244	54	-0.911	0.419

(b) Decay rate: changes between `Variant` at each `LPR`

contrast	LPR	estimate	SE	df	lower.CL	upper.CL
WT - H170A	LPR5	8.451	2.271	52	2.239	14.663
WT - H170D	LPR5	12.699	1.954	52	7.354	18.044
WT - H170E	LPR5	20.319	2.125	52	14.508	26.130
H170A - H170D	LPR5	4.248	2.113	52	-1.530	10.027
H170A - H170E	LPR5	11.868	2.271	52	5.656	18.080
H170D - H170E	LPR5	7.620	1.954	52	2.275	12.965
WT - H170A	LPR10	7.153	3.005	52	-1.065	15.371
WT - H170D	LPR10	13.240	3.005	52	5.022	21.457
WT - H170E	LPR10	19.587	2.271	52	13.375	25.799
H170A - H170D	LPR10	6.087	3.680	52	-3.978	16.151
H170A - H170E	LPR10	12.434	3.110	52	3.928	20.940
H170D - H170E	LPR10	6.347	3.110	52	-2.159	14.853