adding responsive column. see readme
Browse files
README.md
CHANGED
|
@@ -564,6 +564,12 @@ configs:
|
|
| 564 |
dtype: float64
|
| 565 |
description: Benjamini-Hochberg adjusted p-value (DESeq2 output)
|
| 566 |
role: quantitative_measure
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 567 |
|
| 568 |
- config_name: degron_counts_meta
|
| 569 |
description: Sample-level metadata for auxin-inducible degron perturbation experiments with HTSeq count statistics
|
|
|
|
| 564 |
dtype: float64
|
| 565 |
description: Benjamini-Hochberg adjusted p-value (DESeq2 output)
|
| 566 |
role: quantitative_measure
|
| 567 |
+
- name: responsive
|
| 568 |
+
dtype: bool
|
| 569 |
+
description: >-
|
| 570 |
+
TRUE/FALSE labeling based on authors recommended threshold on
|
| 571 |
+
responsiveness where TRUE means that
|
| 572 |
+
padj < 0.1 & abs(log2FoldChange) >= log2(1.3)
|
| 573 |
|
| 574 |
- config_name: degron_counts_meta
|
| 575 |
description: Sample-level metadata for auxin-inducible degron perturbation experiments with HTSeq count statistics
|
rnaseq_reprocessed.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2233d75235f0309d8c03ef92e73ecc1c4e3fdb9cc622bd5b7b910cbf12766e69
|
| 3 |
+
size 41622007
|
scripts/evaluating_zscore_vs_enrichment_vs_peaks_vs_perturbation.R
ADDED
|
@@ -0,0 +1,124 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
library(tidyverse)
|
| 2 |
+
library(here)
|
| 3 |
+
library(ggExtra)
|
| 4 |
+
library(patchwork)
|
| 5 |
+
library(pROC)
|
| 6 |
+
|
| 7 |
+
promoters = read_csv("~/code/hf/yeast_genome_resources/mindel_promoters.csv.gz")
|
| 8 |
+
chec_promoter_sums_raw = read_csv("~/projects/ChEC_Target_Selection/data/chec_sumprom.csv")
|
| 9 |
+
chec_promoter_zscores = chec_promoter_sums_raw %>%
|
| 10 |
+
dplyr::rename(mindel_name = name) %>%
|
| 11 |
+
# removed because these two loci are merged and considered a -1 frameshift
|
| 12 |
+
# of the same locus. confusingly labeled promoters in original mindel data
|
| 13 |
+
filter(!mindel_name %in% c("AAD6", "AAD16", "YAR061W", "YAR062W")) %>%
|
| 14 |
+
pivot_longer(-mindel_name, names_to = 'mindel_regulator', values_to = 'promoter_score') %>%
|
| 15 |
+
left_join(promoters %>%
|
| 16 |
+
select(mindel_name, target_locus_tag, target_symbol)) %>%
|
| 17 |
+
select(-mindel_name) %>%
|
| 18 |
+
left_join(
|
| 19 |
+
promoters %>%
|
| 20 |
+
select(mindel_name, target_locus_tag, target_symbol) %>%
|
| 21 |
+
dplyr::rename(mindel_regulator = mindel_name,
|
| 22 |
+
regulator_locus_tag = target_locus_tag,
|
| 23 |
+
regulator_symbol = target_symbol) %>%
|
| 24 |
+
bind_rows(
|
| 25 |
+
tibble(mindel_regulator = c("COM2", "SDD4"),
|
| 26 |
+
regulator_locus_tag = c("YER130C", "YPR022C"),
|
| 27 |
+
regulator_symbol = c("COM2", "SDD4")))) %>%
|
| 28 |
+
select(-mindel_regulator) %>%
|
| 29 |
+
dplyr::relocate(regulator_locus_tag, regulator_symbol, target_locus_tag, target_symbol) %>%
|
| 30 |
+
group_by(regulator_locus_tag) %>%
|
| 31 |
+
mutate(zscore = as.numeric(scale(promoter_score))) %>%
|
| 32 |
+
ungroup()
|
| 33 |
+
|
| 34 |
+
chec_enrichment = arrow::read_parquet("~/code/hf/mahendrawada_2025/chec_mahendrawada_m2025_af_combined.parquet") %>%
|
| 35 |
+
left_join(arrow::read_parquet("~/code/hf/mahendrawada_2025/chec_mahendrawada_m2025_af_combined_meta.parquet")) %>%
|
| 36 |
+
filter(condition == "standard")
|
| 37 |
+
|
| 38 |
+
mahendrawada_rnaseq = arrow::read_parquet("~/code/hf/mahendrawada_2025/rnaseq_reprocessed.parquet") %>%
|
| 39 |
+
filter(env_condition == "standard_30C") %>%
|
| 40 |
+
replace_na(list(log2FoldChange = 0, padj = 1)) %>%
|
| 41 |
+
mutate(responsive = abs(log2FoldChange) > 0.5 & padj < 0.05)
|
| 42 |
+
|
| 43 |
+
comp_df = chec_promoter_zscores %>%
|
| 44 |
+
select(-c(promoter_score)) %>%
|
| 45 |
+
left_join(select(chec_enrichment, regulator_locus_tag,
|
| 46 |
+
target_locus_tag,
|
| 47 |
+
log_poisson_pval)) %>%
|
| 48 |
+
left_join(select(mahendrawada_rnaseq, regulator_locus_tag, target_locus_tag,
|
| 49 |
+
log2FoldChange, padj, responsive)) %>%
|
| 50 |
+
filter(complete.cases(.)) %>%
|
| 51 |
+
arrange(regulator_locus_tag, target_locus_tag)
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
comp_plot = function(rsym){
|
| 55 |
+
p1 = comp_df %>%
|
| 56 |
+
filter(regulator_symbol == rsym) %>%
|
| 57 |
+
ggplot(aes(zscore, -log_poisson_pval, color = responsive)) +
|
| 58 |
+
geom_point(alpha = 0.8) +
|
| 59 |
+
geom_smooth(aes(group = 1), color = "blue") +
|
| 60 |
+
geom_vline(xintercept = 3.5) +
|
| 61 |
+
geom_hline(yintercept = -log(0.05/5331))
|
| 62 |
+
|
| 63 |
+
df_reg <- comp_df %>% filter(regulator_symbol == rsym)
|
| 64 |
+
|
| 65 |
+
roc_z <- roc(df_reg$responsive, df_reg$zscore, direction = "<")
|
| 66 |
+
roc_p <- roc(df_reg$responsive, -df_reg$log_poisson_pval, direction = "<")
|
| 67 |
+
|
| 68 |
+
roc_plot = ggroc(list(zscore = roc_z, poisson = roc_p)) +
|
| 69 |
+
geom_abline(slope = 1, intercept = 1, linetype = "dashed") +
|
| 70 |
+
scale_color_manual(values = c(zscore = "darkgreen", poisson = "purple")) +
|
| 71 |
+
labs(color = "Method")
|
| 72 |
+
|
| 73 |
+
p2 = comp_df %>%
|
| 74 |
+
filter(regulator_symbol == rsym) %>%
|
| 75 |
+
ggplot(aes(zscore, color = responsive)) +
|
| 76 |
+
stat_ecdf() +
|
| 77 |
+
geom_vline(xintercept = 3.5, linetype = "dashed") +
|
| 78 |
+
labs(y = "Cumulative fraction")
|
| 79 |
+
|
| 80 |
+
p3 = comp_df %>%
|
| 81 |
+
filter(regulator_symbol == rsym) %>%
|
| 82 |
+
ggplot(aes(-log_poisson_pval, color = responsive)) +
|
| 83 |
+
stat_ecdf() +
|
| 84 |
+
geom_vline(xintercept = -log(0.05/5331), linetype = "dashed") +
|
| 85 |
+
labs(y = "Cumulative fraction")
|
| 86 |
+
|
| 87 |
+
(p1 + roc_plot) / (p2 + p3) + plot_layout(guides = "collect") +
|
| 88 |
+
plot_annotation(title = paste0("Regulator: ", rsym))
|
| 89 |
+
}
|
| 90 |
+
|
| 91 |
+
comp_plot("FKH1")
|
| 92 |
+
|
| 93 |
+
comp_auc = comp_df %>%
|
| 94 |
+
filter(!regulator_symbol %in% c("STB5", "USV1")) %>%
|
| 95 |
+
group_by(regulator_symbol) %>%
|
| 96 |
+
reframe(
|
| 97 |
+
auc_zscore = as.numeric(auc(responsive, zscore, direction = "<")),
|
| 98 |
+
auc_poisson = as.numeric(auc(responsive, -log_poisson_pval, direction = "<"))
|
| 99 |
+
)
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
# summary
|
| 103 |
+
comp_auc %>%
|
| 104 |
+
reframe(
|
| 105 |
+
mean_auc_z = mean(auc_zscore),
|
| 106 |
+
mean_auc_p = mean(auc_poisson),
|
| 107 |
+
z_wins = sum(auc_zscore > auc_poisson),
|
| 108 |
+
p_wins = sum(auc_zscore < auc_poisson)
|
| 109 |
+
)
|
| 110 |
+
|
| 111 |
+
comp_df_ranked_by_binding = comp_df %>%
|
| 112 |
+
group_by(regulator_locus_tag) %>%
|
| 113 |
+
mutate(zscore_rank = rank(-zscore, ties.method = "min"),
|
| 114 |
+
enrichment_rank = rank(log_poisson_pval, ties.method = "min")) %>%
|
| 115 |
+
ungroup()
|
| 116 |
+
|
| 117 |
+
rr_25_summary = comp_df_ranked_by_binding %>%
|
| 118 |
+
group_by(regulator_locus_tag) %>%
|
| 119 |
+
reframe(
|
| 120 |
+
zscore_rr_25 = sum(responsive[zscore_rank <= 25]),
|
| 121 |
+
zscore_n = length(responsive[zscore_rank <= 25]),
|
| 122 |
+
enrichment_rr_25 = sum(responsive[enrichment_rank <= 25]),
|
| 123 |
+
enrichment_n = length(responsive[enrichment_rank <= 25])) %>%
|
| 124 |
+
pivot_longer(-regulator_locus_tag)
|