TEGRiP: Transposable Elements Genes RelationshIps Pipeline

TEGRiP is recommended by PCI Genomics 10.24072/pci.genomics.100010 so please cite 10.1101/2021.02.25.432867 if you use it.

Context

The main goal of this project is to find the positional relationships between Transposable Elements (TEs) and genes along genome.

The Apricot genome annotation has been used to validate our strategy (raw data available on ENA PRJEB42606). This pipeline can be used with custom TE annotation as well as de novo assembled genome of any kind of species.

You can learn more about this subject in the Contribution Section.

Introduction

Transposable elements are DNA fragment capable of moving from one place to another troughout the genome via a mecanism called transposition. There are different category/class of transposon. In this project we are going to focus on LTR (long terminal repeat). Learn more by clicking here

Requirement

To use this programme please clone the repository. Make sure that Python3 and Rstudio are installed on your machine. If it's not the case, check out the following links :

• python 3 : link
• Rstudio/R >= 4.0.2 : link

You can create a conda env:

conda create -n tegrip -c "conda-forge/label/cf201901" -c r python==3.7 numpy r-base r-ggplot2 r-optparse
conda activate tegrip

Example

Before running the programme for your own data, please use the testing data to check that everything works.

In the data folder, there are 2 files, one with the data regarding each gene and one regarding each transposon displayed in the diagram above. The python scripts are going to find for each transposable element the nearest gene before and after it. The script will also look for overlapping. The result file of this test data is located in the result folder.

Usage

---

Python Scripts

Input files

The programme takes in consideration the columns name so before using them please verify that each files have those columns :

• Genome files

chromosome	source	feature	start	end	score	strand	phase	ID	Attributes

• TE files

Chromosome	Length_Chr	Type	X	Start	End	X	X	Strand	X	Attribute	X	Class	TE_name	X

• LTR files

species	ID	dfam_target_name	X	X	X	X	chromosome	start	end	strand	X	annotation	X	X	score

Pipeline

Scripts are time optimized using multiprocessing. Multiprocessing is a system that use multiple central processing units (CPUs) making the scripts run faster.

To run the script type the following line by replacing each file name by the real name.

To analyze general TE:

python3 Multiprocessing/Create_Data_TE_multipro_reformatted.py \
	-g data/Gene_testing_data.tsv \
	-te data/Transposon_testing_data.tsv \
	-o result/output_TE.tsv

To analyze LTR:

python3 Multiprocessing/Create_Data_LTR_multiprocessing_reformatted.py \
	-g data/Gene_testing_data.tsv \
	-te data/LTR_testing_data.tsv \
	-o result/output_LTR.tsv

The script will take each file and extract all the data and put them in lists of dictionaries. Then for each TE, it will check the nearest gene whether it's subset,superset, upstream/downstream or whether it's an upstream/downstream overlap.

Output file

The script create a new output file (.tsv) which will be used to make the statistical analysis.

Post analysis

In the first analysis, the strand (+ or -) of genes and TEs are not considered. It can be taken into account by running:

python3 Multiprocessing/Post_TE_multipro_reformatted.py \
	-i result/output_TE.tsv \
	-o result/output_post_TE.tsv

or

python3 Multiprocessing/Post_TE_multipro_reformatted.py \
	-i result/output_LTR.tsv \
	-o result/output_post_LTR.tsv

---

R scripts

There are two R scripts allowing to report different kinds of information. The input must be the result of post-treatment processing to take the strand into account.

• Discriptive stats generates boxplots:
  • with frequency of TEs
  • relationships by TEs
  • quantity of TE by gene
  • quantity of TE by chromosome
• Distance stats shows the number of each TE upstream or downstream with the closest gene (-i -1 -x 0) or within the distance of 0-500 bp (-i 0 -x 500), 500-1000 bp (-i 501 -x 1000), 1000-2000 bp (-i 1001 -x 2000) and more than 2000 bp. The subsets and supersets are not included in these counts.

Rscript Rscript/Descriptive_plot.r \
	-f result/output_post_TE.tsv \
	-o result/TE

Rscript Rscript/Distance_counting.r \
	-f result/output_post_TE.tsv \
	-o result/TE

or

Rscript Rscript/Descriptive_plot.r \
	-f result/output_post_LTR.tsv \
	-o result/LTR

Rscript Rscript/Distance_counting.r \
	-f result/output_post_LTR.tsv \
	-o result/LTR

Contribution

This programme has been developped by Caroline Meguerditchian and Ayse Ergun under the supervision of Marie Lefebvre and Quynh Trang-Bui.

Licence

GNU GENERAL PUBLIC LICENSE