CDR_detect

Detecting CDRs in long reads with methylation tags

|__ scripts
    |__ CDR_detect_reads.py returns readnames that contain potential CDR (older version)
    |__ CDR_detect.py returns readnames and coordinates of CDR in the reads
    |__ evaluate_CDR_detect.py calculates P,R,F1 against input truth CDR reads
    |__ CDR_detect.wdl wdl workflow for hprc assemblies
    |__ Dockerfile used for docker in CDR_detect.wdl
|__ conference_summary.md summary of project July 2022
|__ CDR_detect_notes.md

How to run CDR_detect.py

Python libraries needed:

- pysam
- numpy
- Bio.seq

Command:

python3 CDR_detect.py -b bamfile -o outfile -w <windowsize> -x <methylation threshold> -n <minimum CDR size>

Inputs:

• -b (required) bamfile of reads from the alpha satellite HOR regions. These reads have methylation tags giving the percent likelihood of methylation at each CpG site on the read
• -o (required) output text file name
• -w (default = 1000 bp) Window size of CDRs to consider in bp
• -x (default = 0.3) methylation frequency threshold for calling a CDR in a window
• -n (default = 3000 bp) minimum CDR size in bp

Description of method:

• For each read, move forward in a sliding window of size w by 1 bp
• Calculate methylation frequency in that window
• If the methylation frequency drops below x, record current window start coordinate as start coordinate of a CDR
• Once the methylation frequency rises above x, record current window end coordinate as end coordinate of CDR
• Return readnames and the coordinates of predicted CDRs inside them. Only return CDRs > n (size in bp)