ENST_data_check.Rmd

---
title: "ENST_dataquality_check"
output: html_document
date: "2025-11-20"
---

```{r setup, include=FALSE}
suppressPackageStartupMessages({
  library(data.table)
  library(dplyr)
  library(tidyr)
  library(readr)
  library(glue)
  library(tibble)
  library(stringr)
})

```

```{r}
k2rs <- fread("knownToRefSeq.txt", header = FALSE, sep = "\t",
              col.names = c("kgID","refseq_id")) %>%
  as_tibble() %>%
  mutate(
    kgID      = sub("\\.\\d+$","", kgID),
    refseq_id = sub("\\.\\d+$","", refseq_id)
  ) %>%
  filter(grepl("^(NM|NR|XM|XR)_", refseq_id)) %>%
  distinct()

## ENST → RefSeq (many-to-many)
ens_to_refseq <- known_dt %>%
  dplyr::select(enst = name) %>%
  left_join(k2rs, by = c("enst" = "kgID")) %>%
  filter(!is.na(refseq_id)) %>%
  distinct()

## ------------------------------------------------------------
## 3) Build a RefSeqALL universe from UCSC coordinates
##    (curated refGene + predicted/curated ncbiRefSeq)
##    Schema matches your target: name chrom strand txStart txEnd cdsStart cdsEnd exonCount exonStarts exonEnds score name2 ...
## ------------------------------------------------------------
read_ref_like <- function(path) {
  # refGene/ncbiRefSeq have 16 columns; use friendly names
  fread(path, header = FALSE, sep = "\t", quote = "", na.strings = c("", "NA"),
        col.names = c("bin","name","chrom","strand","txStart","txEnd","cdsStart","cdsEnd",
                      "exonCount","exonStarts","exonEnds","score","name2","cdsStartStat","cdsEndStat","exonFrames")) %>%
    as_tibble() %>%
    mutate(name = sub("\\.\\d+$","", name))
}
refGene_tbl   <- read_ref_like("refGene.txt")
ncbiRefSeq_tbl<- read_ref_like("ncbiRefSeq.txt")

refseq_all <- bind_rows(refGene_tbl, ncbiRefSeq_tbl) %>%
  filter(grepl("^(NM|NR|XM|XR)_", name)) %>%
  arrange(name, chrom, txStart) %>%
  distinct(name, .keep_all = TRUE)
```