Polymorphic Edge Detection (PED) - An efficient polymorphism detector for NGS data

Example of SARS-CoV-2 Mutation Detection

Click to enlarge.

The red vertical line below the s letter of the s gene, which encodes the spike protein around 23 kb in the left figure, is a base substitution mutation of A -> G at the base position of 23403, which is D614G from aspartic acid to glycine.
The graph on the right shows the share of mutant in the total number of patients for each mutation, selecting only the mutations with the outbreak occurred in the United Kingdom from the end of 2020. From end of 2020 to May 2021, lines containing N501Y (A23063T) mutation are majority of SARS-CoV-2 mutations in UK. However the delta line containing L452R (T22916G) invaded the teritory of N501Y line. At June 2021, almost virus are replaced to lines containing the L452R mutation.
The time course is the PED result analyzed for sequence data published by NCBI.

Install PED into linux machine (Ubuntu),
$ sudo apt update
$ sudo apt upgrade
$ sudo apt install curl
$ sudo apt install git
$ git clone https://github.com/akiomiyao/ped.git
Install and setup fastq-dump.
Details for setup is Here.
Download fastq sequence data from NCBI sequence read archive
$ cd ped
$ perl download.pl accession=SRR11513115
$ perl download.pl accession=SRR11513116
$ perl download.pl accession=SRR11513117
$ perl download.pl accession=SRR11513118
$ perl download.pl accession=SRR11542243
$ perl download.pl accession=SRR11542244
Detection of polymorphism
$ perl ped.pl target=SRR11513115,ref=SARS-CoV-2
$ perl ped.pl target=SRR11513116,ref=SARS-CoV-2
$ perl ped.pl target=SRR11513117,ref=SARS-CoV-2
$ perl ped.pl target=SRR11513118,ref=SARS-CoV-2
$ perl ped.pl target=SRR11542243,ref=SARS-CoV-2
$ perl ped.pl target=SRR11542244,ref=SARS-CoV-2
Currently, fastq files of Illumina and Oxford nanopore are supported.
For Docker users
$ sudo apt install docker
$ sudo apt install docker.io
Example of analysis,
$ sudo docker pull akiomiyao/ped
$ sudo docker run -w /ped -v `pwd`:/work akiomiyao/ped perl download.pl accession=SRR11513115,wd=/work
$ sudo docker run -w /ped -v `pwd`:/work akiomiyao/ped perl ped.pl target=SRR11513115,ref=SARS-CoV-2,wd=/work
Copy and paste commands described above to your terminal window, change values of accession, target and run.
Both the ped.pl script and the docker container can be run on Google Cloud Platform and Amazon Web Services.
For private data
$ mkdir your_data1 (target name for ped.pl is the directory name)
$ mkdir your_data1/read
$ cp ~/your_data_directory/foo.fastq your_data1/read (compressed file with gz, bzip2 and xz extentions can be used)
$ perl ped.pl target=your_data1,ref=SARS-CoV-2
or
$ sudo docker run -w /ped -v `pwd`:/work akiomiyao/ped perl ped.pl target=your_data1,ref=SARS-CoV-2,wd=/work
Viewing mutations with Integrative Genomics Viewer
Integrative Genomics Viewer (IVG) is a product of Broad Institute.
Web application of IVG https://igv.org/app/ is useful.
Click to large.
Select Genome to Sars-CoV-2 (ASM985889v3) from the pull down menu.
Download pre-analyzed vcf files linked below.
Import files by selecting 'Local file' in 'Tracks' pull down menu.

If you want to view by your computer, download and install IVG from http://software.broadinstitute.org/software/igv/download.
Click Genomes tab of IVG and load "SARS-CoV-2.fasta" and "SARS-CoV-2.gff" files.
Click File tab and load "ped/SRR11513115/SRR11513115.vcf" and other vcf files or drop files on the main window.

Check mutation in bidirectional alignments
In the ped directory,
$ perl search.pl target=target,chr=chromosome,pos=position

For example,
$ perl search.pl target=SRR11542244,chr=NC_045512.2,pos=5872

or

$ sudo docker run -w /ped -v `pwd`:/work akiomiyao/ped perl search.pl target=SRR11542244,chr=NC_045512.2,pos=5872,wd=/work

A part of output is

# NC_045512.2 5872 NC_045512.2 5870 f deletion -1
CTTACAAAGTCCTCAGAATA GTTACTTATAAATTGGATGG 0
                                                  ChrNC_045512.2 5872
                                                  |
CTTACAAAGTCCTCAGAATACAAAGGTCCTATTACGGATGTTTTCTACAAAGAAAACAGTTACACAACAACCATAAAACCAGTTACTTATAAATTGGATG
||||||||||||||||||||||||||||||||||||||||||||||||||  |||    |     |  | |   ||| |   |   |   || | |   |
CTTACAAAGTCCTCAGAATACAAAGGTCCTATTACGGATGTTTTCTACAAGAAAACAGTTACACAACAACCATAAAACCAGTTACTTATAAATTGGATGG
 |   ||   |     |    || |  |   |   |    |||     ||||||||||||||||||||||||||||||||||||||||||||||||||||
TTACAAAGTCCTCAGAATACAAAGGTCCTATTACGGATGTTTTCTACAAAGAAAACAGTTACACAACAACCATAAAACCAGTTACTTATAAATTGGATGG
                                               |
                                               ChrNC_045512.2 5870

The position for indel in vcf file is one base before the junction position. On the other hand, alignment of ped shows the position of junction.
In the case of no alignment, search with vcf position plus one.

Files
SARS-CoV-2.fasta From NCBI
SARS-CoV-2.gff From NCBI
SRR11542243.count.vcf Sample from University Malaya
SRR11542244.count.vcf Sample from University Malaya
SRR11513115.count.vcf Sample from Utah Public Health Laboratory
SRR11513116.count.vcf Sample from Utah Public Health Laboratory
SRR11513117.count.vcf Sample from Utah Public Health Laboratory
SRR11513118.count.vcf Sample from Utah Public Health Laboratory
Comments
Run time is about 2 min for each accession using Intel Core i5-4210M CPU @ 2.60 GHz (2 cores, 4 threads), 8GB memory.

Reads from Illumina platform returns fine results. On the other hand, results from nanopore contains a lot of false positives.
Higher threshold may be required for analysis of nanopore data.

PED outputs three vcf format files.
'target.vcf' is selected data for IGV.
'target.full.vcf' is full output before the selection.
'target.count.vcf' is from raw data before the verification.
Some mutations have been filtered out by the verification step, because of too much coverage of short reads.
In addition, two mutations in nearby position (within the clipping length) but on different RNA strand (from mixture of viruses) will be filtered out by verification step.
Usage of count.vcf files from raw count data before verification step is recommended.
Acknowledgement
I thank Dr Ilkka Havukkala for testing the PED update for SARS-CoV-2 and advice about integration of IVG.
Citing PED
Cite this article as: Polymorphic edge detection (PED): two efficient methods of polymorphism detection from next-generation sequencing data
Akio Miyao, Jianyu Song Kiyomiya, Keiko Iida, Koji Doi, Hiroshi Yasue
BMC Bioinformatics. 2019 20(1):362.
URL https://doi.org/10.1186/s12859-019-2955-6
Author
MIYAO Akio (miyao@affrc.go.jp)