In kissmyjazz/ketamineCoxPaper: Code for the paper on ketamine-COX experiment

Testing for difference in mean COX levels

fp_table <- here::here("analysis", "data", "derived_data", "t_tests.rds")
# remove Cpu_background...1.3.
t_tests <- readr::read_rds(fp_table) %>% slice(-9) %>% dplyr::mutate(diff = papaja::printnum(`mean difference`)) %>% dplyr::select(-(5:8)) %>% dplyr::select(1, starts_with("diff"), everything()) 
colnames(t_tests) <- c("\nBrain region", "difference of the means vs. control", "\nt", "\ndf", "\np")
t_tests %>%
kable(escape = T, booktabs = T, format = "latex", align = rep("l", 5),
      digits = 3, caption = "Brain regions with increased COX levels ($\\mu$mol/min/gram tissue) after subchronic ketamine treatment") %>%
  kable_styling(bootstrap_options = "striped", 
                latex_options="scale_down", font_size = 12) %>% 
  column_spec(1, "18em") %>% column_spec(2, "8em")

After we had performed the Welch's two sample t-test on 10 multiply imputed datasets and pooled the estimates, the mean comparisons in r nrow(t_tests) brain regions, mostly those instrumental to perception of the environment and/or implicated in schizophrenia, emerged as statistically significant. The results of t-test comparisons are provided in Table \@ref(tab:table-t-tests). In all significant comparisons ketamine-treated rats had higher levels of COX activity.

Differential correlation analysis

fp_overall_pvalue <- here::here("analysis", "data", "derived_data", "dgca_imputed_overall_pvalue.rds")
p_value <- readr::read_rds(fp_overall_pvalue) 

fp_avg_zscore <- here::here("analysis", "data", "derived_data", "dgca_imputed_ave_zscore.rds")
sign_cor_changes <- readr::read_rds(fp_avg_zscore)
sign_cor_changes %>% dplyr::rename("Brain region" = regions, "\u0394 z-score" = z_score_diff) %>% papaja::apa_table(escape = FALSE, digits = 3, caption = "Regional median changes in correlation coefficients")

The median correlation between the two conditions did not differ significantly p = r papaja::printnum(p_value). The regional median changes in correlations are provided in Table \@ref(tab:dca-regional-average). Five brain regions had significantly more negative median correlations with the rest of brain regions in ketamine-treated rats. The locus coeruleus instead showed a borderline significant increase in the median correlation.

fp_pos_df <- here("analysis", "data", "derived_data", "control_pos_df.rds")
fp_neg_df <- here("analysis", "data", "derived_data", "control_neg_df.rds")
control_pos_df <- readr::read_rds(fp_pos_df)
control_neg_df <- readr::read_rds(fp_neg_df)

control_pos_df %>% dplyr::rename("region 1" = region1,
  "region 2" = region2, "\u0394 z-score" = z_score_diff, "r control" = control_cor, 
  "r ketamine" = treatment_cor) %>% 
  kable(booktabs = T, format = "latex", longtable = TRUE, digits = 3, 
      caption = "Decreased pairwise correlations in ketamine rats") %>%
kable_styling(bootstrap_options = "striped", font_size = 12, latex_options = "repeat_header")

control_neg_df %>% dplyr::rename("region 1" = region1,
  "region 2" = region2, "\u0394 z-score" = z_score_diff, "r control" = control_cor, 
  "r ketamine" = treatment_cor) %>% 
  kable(booktabs = T, format = "latex", longtable = TRUE, digits = 3, 
      caption = "Increased pairwise correlations in ketamine rats") %>%
kable_styling(bootstrap_options = "striped", font_size = 12)

fp_negative_pairs <- here("analysis", "data", "derived_data", "negative_pairs.rds")
fp_positive_pairs <- here("analysis", "data", "derived_data", "positive_pairs.rds")
names_pos <- readr::read_rds(fp_positive_pairs)
names_neg <- readr::read_rds(fp_negative_pairs)

(ref:corplot) Graphical representation of changes in correlation coefficients after subchronic ketamine treatment. Line segments connect pairs of brain regions with significant difference in correlation scores between the two conditions. The widths of line segments correspond to the magnitude of the difference between two correlation scores.

g_graph_fp <- here("analysis", "data", "derived_data", "g_graph.rds")
g <- readr::read_rds(g_graph_fp)
cor_plot <- g + geom_edge_link(aes(color = direction, width = abs(z_score_diff)),
                 check_overlap = TRUE) +
  geom_node_point(aes(color = as.factor(nomenclature)), size = 3) +
  geom_node_text(aes(label = name), vjust = -0.5, hjust = 0.5, family = "serif") +
  scale_edge_color_brewer(palette = "Pastel1") +
  scale_color_brewer(palette = "Set1") + labs(
    title = "Changes in correlation scores after subchronic ketamine treatment",
    color = "Brain structures:",
    edge_color = "Change in pairwise correlations vs. control group:",
    edge_width = "Delta z-score:", parse = TRUE) +
  theme_void() + theme(legend.position = "bottom",
                       legend.box = "vertical",
                       plot.margin=unit(c(0.5,1,1,1.1),"cm"),
                       plot.title = element_text(hjust = 0.5, size = 12, face = "bold",
                                                 margin=margin(0,0,20,0)),
                       legend.text = element_text(size=rel(0.9)),
                       legend.key.width = unit(0.4, "cm"),
                       legend.key.height = unit(0.5, "cm")) +
  guides(edge_color = guide_legend(override.aes = list(edge_width = 3, linetype = 0)))
cor_plot

Table 3 presents pairwise correlations at p level below 0.01. In the majority of cases (r nrow(control_pos_df)), the positive association in control rats was abolished (high positive correlation either became non-significant or negative) in ketamine-treated rats. Optic nerve layer of the superior colliculi was the brain region with the biggest number of significant pairwise changes in correlation (8 pairs). Other brain regions with significant changes in connectivity had between 2 and 4 pairwise reductions in the correlation coefficients: anterior pretectal nucleus; hippocampal regions CA2, CA3, and dentate gyrus; dorsal lateral geniculate nucleus; dorsomedial periaqueductal gray matter; ectorhinal cortex; superficial gray, intermediate gray and white matter layers of the superior colliculi; prerubral field; granular retrosplenial cortex, secondary visual cortex (area V2). Conversely, the opposite pattern was observed in r nrow(control_neg_df) pairs, where the correlation increased after ketamine administration (Table \@ref(tab:table-negative)). Here, the brain regions with 2 pairwise reversals in the direction of correlation were as follows: anterior cingulate cortex, ventral medial geniculate nucleus, medial mamillary bodies, and reticulotegmental nucleus. The pairwise changes in correlation coefficients after subchronic administration of ketamine are summarised in Figure \@ref(fig:cor-plot). Full results at p level below 0.05 are presented in Figure 2 on the Supplementary materials. In total, 336 pairwise comparisons were significant at p level below 0.05, which constitutes 1.9% of all possible pairs. In 66% of the cases ketamine-treated rats had significantly less positive pairwise correlations than control group animals.

kissmyjazz/ketamineCoxPaper documentation built on Nov. 4, 2019, 3:51 p.m.