From 3a10380d8c461f4bf38655e2008bf55aa648da1a Mon Sep 17 00:00:00 2001
From: coolneng <akasroua@gmail.com>
Date: Thu, 25 Mar 2021 21:53:49 +0100
Subject: [PATCH] Construct a dataframe containing the HVR region

---
 src/alignment.r | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/src/alignment.r b/src/alignment.r
index ea6d966..a0a2d57 100644
--- a/src/alignment.r
+++ b/src/alignment.r
@@ -1,12 +1,23 @@
 library(Biostrings)
-library(fastqcr)
+library(parallel)
+
+construct_dataframe <- function(data) {
+  vdj_string_set <- lapply(data, FUN = Biostrings::DNAStringSet)
+  vdj_dataframe <- as.data.frame(vdj_string_set)
+  vdj_dataframe$hvr_region <- paste(vdj_dataframe$v_sequence,
+    vdj_dataframe$d_sequence, vdj_dataframe$j_sequence,
+    sep = ""
+  )
+  return(vdj_dataframe)
+}
 
 parse_data <- function(files) {
   reversed_sequences <- Biostrings::readQualityScaledDNAStringSet(files[1])
   sequences <- Biostrings::reverseComplement(reversed_sequences)
   vdj_alignment <- read.csv(files[2])
-  vdj_string_set <- lapply(vdj_alignment, FUN = Biostrings::DNAStringSet)
-  return(list(sequences, vdj_string_set))
+  vdj_dataframe <- construct_dataframe(vdj_alignment)
+  return(list(sequences, vdj_dataframe))
+}
 }
 
 align_sequences <- function(sequences, vdj_segments) {