egfr-v3-determiner
EGFRvIII is the most common mutation found in the EGFR gene in glioblastomma. It originates from a genomic deletion or multiple inversion resulting in the loss of exons 2 - 7 at RNA level.
Using RNA-seq, EGFRvIII can be determined based on reads that:
- (1) fall exactly over the splice junction (use
-s/--spliced-only) - (2) are 'spanning' the splice junction and are thus mapped perfectly within exons 1 and 8
The only thing that you need are properly aligned RNA-seq BAM files and this tool.
Please notice that type 2 reads can also arise with other structural variants in combination with large insert sizes. For instance, if exons 2 - 6 are deleted and the length of the inner sequence of the RNA frament is longer than exon 7, it could be wrongly interpreted as EGFRvIII read.
What is does
We designed a small python tool for estimnating the read counts and/or extracting the actual read names that allows to further analysed the sequencing data.
Estimates the number of EGFR-vIII and EGFR-wt reads from BAM files directly:
$ egfr-v3-determiner -r hg38 tmp/test_001.bam tmp/test_002_wt_non-spliced.bam
Will result in a text file like this:
| sample | wt-reads | vIII-reads |
|---|---|---|
| tmp/test_001.bam | 0 | 1 |
egfr-v3-determiner Manuscript
If you like to read more about the application of egfr-v3-determiner or when using this package for scientific purposes, please read/cite the following manuscript:
Youri Hoogstrate, Santoesha A Ghisai, Maurice de Wit, Iris de Heer,
Kaspar Draaisma, Job van Riet, Harmen J G van de Werken, Vincent Bours,
Jan Buter, Isabelle Vanden Bempt, Marica Eoli, Enrico Franceschi,
Jean-Sebastien Frenel, Thierry Gorlia, Monique C Hanse, Ann Hoeben,
Melissa Kerkhof, Johan M Kros, Sieger Leenstra, Giuseppe Lombardi,
Slávka Lukacova, Pierre A Robe, Juan M Sepulveda, Walter Taal,
Martin Taphoorn, René M Vernhout, Annemiek M E Walenkamp, Colin Watts,
Michael Weller, Filip Y F de Vos, Guido W Jenster, Martin van den Bent,
Pim J French.
The EGFRvIII transcriptome in glioblastoma, a meta-omics analysis,
Neuro-Oncology, 2021; noab231,
https://doi.org/10.1093/neuonc/noab231
This will indirectly help me further maintaining this software package.
Thank you in advance, Youri
Installation
git clone https://github.com/yhoogstrate/egfr-v3-determiner.git
cd egfr-v3-determiner
virtualenv -p python3 .venv
source .venv/bin/activate
python setup.py install
egfr-v3-determiner --help
nosetests tests/*.py
Usage
Usage: egfr-v3-determiner [OPTIONS] [INPUT_BAM]...
Options:
--version Show the version and exit.
-r, --reference-build [hg19|hg38]
Used reference genome (needed for EGFR exon
coordinates) [required]
-s, --spliced-reads-only If paired end reads with an insert size
longer than 801 bases can be expected, wild-
type exon-1 to exon-8 covering reads can can
be expected. Enabling this flag only uses
spliced reads for vIII determination.
-n, --read-names Report all read-names instead of the read
counts.
-i, --include-interchromosomal Include paired-end reads that have an
interchromosomal mapped mate (disabled by
default).
-d, --dataset-suffix TEXT Adds this suffix to the column names; tabs
and newlines not allowed.
-f, --include-duplicates Force including duplicate reads (as marked
by samtools/sambamba/picard etc. - disabled
by default).
-w, --exons-wt TEXT Exons included for counting (--spliced-
reads-only disabled). Options: '2' '2,3'
(equals 'default'), '2,3,4', '2,3,4,5',
'2,3,4,5,6', '2,3,4,5,6,7' (equals 'all'),
default: '2,3'.
-v, --exons-viii TEXT Exons included for counting (--spliced-
reads-only disabled). Options: '8' '8,9'
(equals 'default'), '8,9,10' (equals 'all'),
default: '8,9'.
--help Show this message and exit.
Genomic reference
The genomic reference (hg19/hg39) can be changed using the -r or
--reference-build argument. Genomic references starting with '>1' rather
than 'chr1' will be automatically resolved.
Interchromosomal reads
It may happen that one of the mates splices perfectly over exons 1 - 8 but that it's mate is mapped to another chromosome. As these are likely derived from other stuctural variants or odd reads, they are by default excluded.
It may nevertheless, be interesting to analyse these reads, for instance if
there is a suspicion for other structural variants. By using the -i /
--include-interchromosomal argument, these reads will be included.
Suffix output column names
It may be convenient to add a suffix to the column name in the output, for
instance '-EGFRvIII-reads'. This can be achieved by using the -d /
--dataset-suffix argument:
egfr-v3-determiner -d '-EGFRvIII-reads' sample-1.bam sample-2.bam > EGFRvIII.cnts.txt
I NEED HELP / I FOUND A BUG / I WANT A FEATURE
Great, please do so :) I am more than happy to help. If you find some odd reads that I need to take a look at, feel free to send them to me.
If you prefer private communication you can e-mail me at:
y {dot} hoogstrate {at} erasmusmc {.} nl
Duplicate reads
Reads marked as duplicate (using sambamba / Picard / samtools etc.) are by default not included for counting. This is because they typically account for more data points but not for more information. You can compare it with scaling a bitmap figure; it contains more pixels but not more information.
If the data is single-end sequenced to an extreme depth, you may like to include PCR duplicates as they could be coincidentallly marked as duplicate while they originate from actual distinct RNA molecules. One may wonder why duplicate marking was performed in the first place.
Nevertheless, to enable countingfor duplicate reads, use -f /
--include-duplicates.
Modify used exons
The used EGFR exons can be changed using the following arguments:
-w, --exons-wt TEXT Exons included for counting (--spliced-
reads-only disabled). Options: '2' '2,3'
(equals 'default'), '2,3,4', '2,3,4,5',
'2,3,4,5,6', '2,3,4,5,6,7' (equals 'all'),
default: '2,3'.
-v, --exons-viii TEXT Exons included for counting (--spliced-
reads-only disabled). Options: '8' '8,9'
(equals 'default'), '8,9,10' (equals 'all'),
default: '8,9'.
Ths only works when non-spliced alignments are permitted.
LICENSE
This is FREE software without liability or warranty, following the GPL-3 software license.