index_Thomas.Rmd

---
title: "eQTL Website"
---

New example!!!!

```{r init, include=FALSE}
library(tidyverse)
# import package to allow R data objects to be displayed as tables on HTML pages
library(DT)
```


## Regular Gene-SNP table

```{r include=FALSE}
# init. filepath
path_to_genes <- "website/data/plots/"
# init. genes names to store in DF
genes <- list.files(path_to_genes)

# init. DF with gene and SNP combos
df <- tibble(
  gene = NA,
  snp = NA
)

for (gene in genes) {
  # init. vector of SNPs for each gene
  snps <- list.files(str_c(path_to_genes, gene))
  
  # init. temporary DF to store each gene-SNP combination
  temp_df <- tibble(gene = gene, snp = snps)
  
  # bind DF with temporary DF
  df <- bind_rows(df, temp_df)
}

df <- df %>% filter(!is.na(gene))

# df <- df %>%
#   filter(!is.na(gene)) %>%
#   datatable(escape = F, style = "auto")
```



```{r display table, echo=FALSE, include=FALSE}
df_html <- df %>%
  datatable(escape = F, style = "auto")

df_html
```

---

## LD Trait Association

TODO: 

* load this table in a separate tab
* make a dropdown menu for `GWAS_Trait` 
* select specific SNPs to view (from `Query`) -- optional for now
* make `R2` filterable (e.g., view everything between x and y)
* make hyperlink from the Feature-Selected table that shows the LDTrait records for the specific SNP (`Query`)
    - filter for the SNP

```{r}
ld_trait_assoc <- read_tsv("./website/data/modeling_results/QTLs_LDtrait_association_ALLpopulation_r20.1.txt")
```




---

## Feature-Selected Model: Phenotype Rank-Normalized CPM

To-Do:

* convert numeric cols to scientific notation
* link the GTEx IGV browser to the table ("location=....")


```{r include=FALSE}
# read in feature selected rank-normalized CPM model
fs_rank_norm_cpm <- read_tsv("website/data/modeling_results/modeling_results_featureSelectedModel_phenotypeRankNormCPM.txt")


fs_rank_norm_cpm <- fs_rank_norm_cpm %>%
  left_join(
    df,
    by = c("GENE" = "gene", "SNP" = "snp")
  )
```


```{r echo=FALSE, warning=FALSE, message=FALSE}
# init. URL base
url_gene_base <- "https://www.genecards.org/cgi-bin/carddisp.pl?gene="
url_snp_base <- "https://www.ncbi.nlm.nih.gov/snp/"

# embed hyperlinks for each SNP and gene
mod_1 <- fs_rank_norm_cpm %>%
  mutate(
    gene = str_c("<a href=", url_gene_base, GENE, " target=_blank", ">", GENE, "</a>"),
    snp = str_c("<a href=", url_snp_base, SNP, " target=_blank", ">", SNP, "</a>")
  )

# create a `plots` column to store the plots
# mod_1 %>%
#   knitr::kable(col.names = colnames(mod_1)) %>%
#   mutate(plots = str_c("![Paired Plot CPM]('website/data/plots/", GENE, SNP, "pairedPlot_3donorColored_CPM.png')", sep = "/"))

fs_rank_norm_cpm %>%
  mutate(
    gene = str_c("<a href=", url_gene_base, GENE, " target=_blank", ">", GENE, "</a>"),
    snp = str_c("<a href=", url_snp_base, SNP, " target=_blank", ">", SNP, "</a>"),
    # plots = str_c("![Paired Plot CPM](website/data/plots", GENE, SNP, "pairedPlot_3donorColored_CPM.png)", sep = "/"),
    # plots = str_c("<img src=website/data/plots/", GENE, SNP, "pairedPlot_3donorColored_CPM.png target=_blank>Paired Plot CPM</img>", sep = "/"),
    plots = str_c("<a href=website/data/plots/", GENE, "/", SNP, "/", "pairedPlot_3donorColored_CPM.png target=_blank>Paired Plot CPM</a>"),
    Actions = str_c("LDlink", "IGV Browser", sep = ", "),
    # convert specific cols to sci notation
    across(.cols = str_subset(colnames(.), "(pval|pperm|beta|_se)"),
           .fns = ~ format(as.numeric(.x), scientific = TRUE, digits = 3))
  ) %>%
  select(pair, snp, gene, plots, Actions, everything(), -c(GENE, SNP)) %>%
  datatable(escape = F, style = "auto")

# options(scipen = 3)
```


![Paired Plot CPM](./website/data/plots/THUMPD2/rs6759076/pairedPlot_3donorColored_CPM.png)


reate mode 100644 website/example/ERAP2/rs10038651/ERAP2_rs10038651.html
 create mode 100644 website/example/ERAP2/rs10038651/FRB_plot.pdf
 create mode 100644 website/example/ERAP2/rs10038651/KRT_plot.pdf
 create mode 100644 website/example/ERAP2/rs10038651/MEL_plot.pdf
 create mode 100644 website/example/ERAP2/rs10038651/modeling_result_FRB.txt
 create mode 100644 website/example/ERAP2/rs10038651/modeling_result_KRT.txt
 create mode 100644 website/example/ERAP2/rs10038651/modeling_result_MEL.txt
 create mode 100644 website/example/ERAP2/snps.txt
UMMSMAC-G7JPHWFY6Y:eQTL_website.github.io tjacob$ 
UMMSMAC-G7JPHWFY6Y:eQTL_website.github.io tjacob$ git push -u origin HEAD:main
Enter passphrase for key '/Users/tjacob/.ssh/id_ecdsa': 
Enumerating objects: 79850, done.
Counting objects: 100% (79850/79850), done.
Delta compression using up to 12 threads
Compressing objects: 100% (67563/67563), done.
remote: fatal: pack exceeds maximum allowed size (2.00 GiB)
error: remote unpack failed: index-pack failed| 8.91 MiB/s
To github.com:garber-lab/eQTL_website.github.io.git
 ! [remote rejected]     HEAD -> main (failed)
error: failed to push some refs to 'github.com:garber-lab/eQTL_website.github.io.git'


```{r}
list.files("website/data/plots/") %>% str_subset("^I")
```




---


## Feature-Selected Model: Phenotype CPM

```{r feature_selected_models, include=FALSE, eval=FALSE}
# read in different modeling results as tables
feat_select_CPM <- read_tsv("website/data/modeling_results/modeling_results_featureSelectedModel_phenotypeCPM.txt")

feat_select_CPM <- feat_select_CPM %>%
  left_join(
    df,
    by = c("GENE" = "gene", "SNP" = "snp")
  )
```


```{r include=FALSE, eval=FALSE}
# init. URL base
url_gene_base <- "https://www.ncbi.nlm.nih.gov/gene/?term="
url_snp_base <- "https://www.ncbi.nlm.nih.gov/snp/"

feat_select_CPM %>%
  mutate(
    GENE = str_c("<a href=", url_gene_base, GENE, " target=_blank", ">", GENE, "</a>"),
    SNP = str_c("<a href=", url_snp_base, SNP, " target=_blank", ">", SNP, "</a>")
  ) %>%
  datatable(escape = F, style = "auto")
```