NGI-AS-tech_note-2024

Introduction

This repository contains supplementary files for NGI tech note "Tech Note - Adaptive Sampling: targeted Oxford Nanopore long-read sequencing" published in 2024.

BED files

There are two sets of BED files with targets. One in hg38 co-ordinate space (in bedfiles_hg38/) and another one that requires a custom reference genome (found in bedfiles_customref).

In brief, the way to recreate the custom reference files is to use BEDTools getfasta on hg38 to extract the ROIs to ROI.fasta (rename the fasta entries to "ROI"), then use bedtools maskfasta to softmask the ROI positions to hg38_masked.fasta Then cat ROI.fasta hg38_masked.fasta > custom_hg38.fasta The reason for using this custom reference method is to make it easier to validate in real-time on the instrument that enrichment is occuring:

experiment_sample	hg38_bed	custom_bed
P29602_01pct	day1_0.1pct_targets.bed	day1_0.1pct.bed
P29602_025pct	day1_0.25pct_targets.bed	day1_0.25pct.bed
P29602_05pct	day1_0.5pct_targets.bed	day1_0.5pct.bed
P29602_1pct	day1_1pct_targets.bed	day1_1pct.bed
P29602_4pct	day1_4pct_targets.bed	day1_4pct.bed
P29602_8pct	day1_8pct_targets.bed	day1_8pct.bed
P29602_ROI_as_ref	day1_ROI_as_ref_1pct_targets.bed	day1_ROI_as_ref_1pct.bed
P29602_01pct_10ROIs	1.bed	-
P29602_01pct_80ROIs	2.bed	-
P29602_day3_gapfill	day3_nomod_targets.bed	day3_gapfill.bed
Reserved channel experiment	day1_8pct_targets.bed	day1_8pct.bed

Command-line scripts

This is a collection of bespoke one-liners for processing the data in the tech note

Split channels in the reserved channel experiment

# bioawk: https://github.com/lh3/bioawk
# Channels 1-1499 with adaptive sampling
bioawk -cfastx 'match($comment, /ch=[[:digit:]]+ /){if(substr($comment, RSTART+3, RLENGTH-3) < 1500){print "@"$name" "$comment"\n"$seq"\n+\n"$qual}}' passed.fastq.gz | gzip -c > AS.fastq.gz
# Channels 1500-3000 without AS
bioawk -cfastx 'match($comment, /ch=[[:digit:]]+ /){if(substr($comment, RSTART+3, RLENGTH-3) >= 1500){print "@"$name" "$comment"\n"$seq"\n+\n"$qual}}' passed.fastq.gz | gzip -c > noAS.fastq.gz

Count active channels per time interval (fig 5)

# Method: Bin reads (passed + failed) into start_time hour. Find and count every 
# unique channel per bin. For future reference: just use nanoplot for this.

bioawk -cfastx '{print $comment}' AS_failed_and_passed.fastq.gz | grep -o "ch=\w* start_time=2023-\w\w-.*+02:00" | cut -b 4- | sed 's/ start_time=/ \t/' | grep -o ".*2023-09-[0-9]*T[0-9]*" | awk '{print $2":00:00\t"$1}' | sort | uniq | cut -f 1 | uniq -c | awk '{print $2"\t"$1}' > AS_chpr.txt