TEMP2 is an algorithm for detecting transposon insertions using short-read whole-genome sequencing data. It can not only precisely detect germline transposon insertions, but also estimate the number of uninherited/somatic transposon insertions by removing artificial insertion introduced by chimeric reads.
If you use TEMP2 for transposon insertion detection, please cite:
Yu et al. A benchmark and an algorithm for detecting germline transposon insertions and measuring de novo transposon insertion frequencies. Nucleic Acid Research. 2021.
Current version: v0.1.4
- Linux x86_64 systems.
- Perl package BioPerl is needed for running absence module.
- Although other third-party tools have already pre-compiled and included with TEMP2, you may want to have them installed in your environment to avoid dependency errors.
With git installed, simply type the following command to install TEMP2:
git clone https://github.com/weng-lab/TEMP2 ln -s $PWD/TEMP2/TEMP2 your_bin_path/
To avoid mixing the pre-compiled tools with your own versions, we do not recommend to add
/TEMP2/bin to the
Alternatively, you can also install TEMP2 after fetching source code:
tar -xzvf TEMP2.tar.gz cd TEMP2 ln -s $PWD/TEMP2 your_bin_path/
TEMP2 integrates a tested dataset in /TEMP2/test/
To test if TEMP2 is successfully installed, you can run the command below:
cd TEMP2/test TEMP2 insertion -l test.1.fastq.gz -r test.2.fastq.gz -I bwa_index/chr2L -g chr2L.fa -R transposon.fa -t rmsk.bed -o test_output -c 2
This command takes around 2 minutes with 2 avaiable CPU. A successul installation should outputs
test.soma.summary.txt in the folder
test_output, which contains detected insertions and estimated de novo insertion number per genome respectively.
TEMP2 includes three modules:
- insertion. The most general module that detects both germline and de novo transposon insertions.
- insertion2. A simplified module that only detects germline transposon insertions.
- absence. The module annotates the absence of reference transposon copies. Same as
TEMP_absence.shin our previous version ––– TEMP.
After installation, typing
TEMP2in the terminal produces a list of usage options, which are:
TEMP2 -- Version 0.1.1 -- germline and de novo transposon insertion and deletion analysis usage: TEMP2 insertion -h # for germline and somatic transposon insertion TEMP2 insertion2 -h # for only germline transposon insertion TEMP2 absence -h # for transposon absence or deletion
Detect germline and de novo transposon insertions
To see the help information of insertion module, type
TEMP2 insertion in the terminal.
Germline and somatic transposon insertion detection using short DNA-seq data. Please send questions, suggestions and bug reports to firstname.lastname@example.org. Thank you for using it. ============================================================================= usage: TEMP2 insertion -l read1.fq Read1 for paired-end DNA-seq, can be gzipped. -r read2.fq Read2 for paired-end DNA-seq, can be gzipped. -i map.bam Genome mapping file in sorted and indexed bam format. Use one of -i or -I -I bwa_index You can also input bwa_index instead of input map.bam, TEMP2 can map DNA-seq to the genome. -g genome.fa Genome sequences in fasta format. -R TE.fa Transposon consensus sequences in fasta format. -t TE.bed6 Transposon copies annotated in genome in bed6 format. You can download it from UCSC or creat it by running RepeatMasker. Options: -o out_path Output directory for results. Default is assigned by the prefix of read1.fq. -p out_prefix Prefix to output directory. Default is assigned by the prefix of read1.fq. -A Set this parameter to enable ALU mode, which allows the insertion between two concordantly mapped reads -M mismatch% Percentage of mismatch allowed when anchor to genome. Default is 2. -m mismatch% Percentage of mismatch allowed when mapping to TEs. Default is 5. -U ratio The ratio between the second best alignment and the best alignment to judge if a read is uniquely mapped. Default is 0.8. -f frag_length Fragment length of the library. Default is calculated based on the mapping result. -N reference_filter_window window sizea (+-n) for filtering insertions overlapping reference insertions. Default is 300. -G number_of_genome Suggested when you are sequencing limited genomes in your library.i TEMP2 need this to set a reasonable cutoff to define potential de novo insertions. -C frequency_cutoff Lower than which frequency should TEMP2 regard a insertion as poteintial de novo insertion. By default TEMP2 uses singleton insertions (1 supporting read) as de novo insertions becasue usually the sequencing depth is far less than number of genomes in the library. If -G is set, TEMP2 uses 2-fold of theoretically frequency of de novo insertions as cutoff. However,you can always set a self-defined frequency cutoff to seprate de novo and germline insertions. -T Set this parameter to allow truncated de novo insertions; For default, only full-length de novo insertions are allowed. -L Set this parameter to use a looser criteria to filter reference annotated copy overlapped insertions; Default not allowed. -S Set this parameter to perform insertion length checking; Default is to reserve those insertions that are not full length of shorter than 500bp. -c cpu_number Number of CPU used. Default is 1. -d Set this parameter to delete tmp files. Default is moving them to folder tmpTEMP2. -h Show this message. -v Show version information.
The insertion2 module accepts exactly the same arguments as insertion. However, you can also type
TEMP2 insertion2 to get the detailed help information.
TEMP2 insertion outputs several files. Assuming the output prefix is test, then the following files shall be present in the ouput directory:
test.insertion.bed: a bed format file with additional 8 columns (6+8).
Column 1,2&3: Reference genome position (chromosome, start, and end) of the transposon insertion.
Column 4: Information of inserted transposon, including transposon name, start coordinate and end coordinate of the inserted transposon, and which strand it inserts. Separated by
Column 5: Frequency of the inserted transposon. It generally means what fraction of sequenced genome present this insertion.
Column 6: Which strand this transposon inserts.
Column 7: Type of the insertion. We typically separate insertions into three categories based on how many reads support them.
1p1: Insertions supported by reads at both ends.
2p: Insertions supported by multipe reads at only one end.
singleton: Insertions supported by only one read.
Ps: We used to regard 1p1 insertions as confident germline insertions but are abondon this filtering creteria now. The reason is that our recent study found supporting reads showed bias in the two ends of some transposons, especially LINE elements in fruit fly.
Column 8: Number of reads supporting this insertion.
Ps: If you are annotate confident transposon insertions, make sure to filter insertions without enough supporting reads, as many of them are false positives.
Column 9: Number of reads that do not support this insertion, AKA reference reads.
Column 10: Number of supporting reads at 5'end of the insertion.
Column 11: Number of supporting reads at 3'end of the insertion.
Column 12: Target site duplication (TSD) of the insertion. unknown is shown if not applicable.
Column 13: Reliability of this insertion (0–100). 100 for 2p and 1p1 insertions. For singleton insertions, TEMP2 already filtered out most of the false positives but not all of them. The reliability is a percentage stand for how many singleton insertions of a specific transposon is.
Column 14: Number of supporting reads at 5'end of the insertion junction.
Column 15: Number of supporting reads at 3'end of the insertion junction.
test.soma.summary.txt: a tab delimited file includes 10 columns (not included when running insertion2 module)
Column 1: Name of transposon.
Column 2: Estimated number of de novo insertion of this transposon per genome.
Column 3: 95th percentile (lambda distribution) number of de novo insertion of this transposon per genome.
Column 4: Total estimated number of de novo insertion of this transposon.
Column 5: 95th percentile (lambda distribution) number of de novo insertion of this transposon.
Column 6: Number of singleton reads mapped to the end regions (positive regions) of this transposon.
Column 7: Number of singleton reads mapped to the center region (negative region) of this transposon.
Column 8: Number of reads mapped to the end regions (positive regions) of this transposon.
Column 9: Number of reads mapped to the center region (negative region) of this transposon.
Column 10: Status of the estimation.
test.supportReadsUnfiltered.bb: bigBed6 file for all supporting reads.
test.supportingRead.dis.pdf: figures showing where the supporting reads mapped to each transposon.
tmpTEMP2: folder contains all the intermediate files. eg: you can find fragment length statistics in tmpTEMP2/test.fragL
Detect transposon absense in the reference genome
This module renders the same code as TEMP. Typing
TEMP2 absence shows the help information:
usage: ./TEMP2_absence.sh -i input_file.sorted.bam -s scripts_directory -o output_directory -r transposon_rpmk.bed -t reference.2bit -f fragment_size -c CPUs -h TEMP is a software package for detecting transposable elements (TEs) insertions and excisions from pooled high-throughput sequencing data. Please send questions, suggestions and bug reports to: email@example.com Options: -i Input file in bam format with full path. Please sort and index the file before calling this program. Sorting and indexing can be done by 'samtools sort' and 'samtools index' -s Directory where all the scripts are -o Path to output directory. Default is current directory -r Annotated transposon positions in the genome (e.g., repeakMask) in bed6 format with full path -t 2bit file for the reference genome (can be downloaded from UCSC Genome Browser) -f An integer specifying the length of the fragments (inserts) of the library. Default is 500 -x The minimum score difference between the best hit and the second best hit for considering a read as uniquely mapped. For BWA MEM. -c An integer specifying the number of CPUs used. Default is 4 -h Show help message
For transposon absence analysis, the summay output file remains exactly the same as TEMP.
- test.absence.refined.bp.summary: There are 9 columns in the summary file and their meanings are listed below:
Column 1,2&3: Reference genome position (chromosome, start, and end) of the transposon absence.
Column 4: The TE family that the detected insertion belongs to.
Column 5: Junctions at 5’ of the excised TE. The two numbers are the coordinates of the junctions on the two strands.
Column 6: Junctions at 3’ of the excised TE. The two numbers are the coordinates of the junctions on the two strands.
Column 7: The number of reads supporting the absence.
Column 8: The number of reads supporting the reference (no absence).
Column 9: Estimated population frequency of the detected absence event.
- Add -G for users to input number of genomes used in the library and TEMP2 can automatically choose threshold to detect potential de novo insertions.
- Change -C from self-defined read cutoff to self-defined frequency cutoff.
- By default, TEMP2 hypothesize that de novo insertions that are derived during individual development is full-length. However, a few transposons does have truncated insertions, such as 5' truncated L1 elements. Now, users can use the option -T to allow truncated de novo insertions. In this option, TEMP2 will firstly mark those truncated insertions that are supported by at least 3 reads at both ends. Then marked truncated insertions together with full-length insertions are used by TEMP2 for estimating de novo insertion numbers.
- Usually, Illumina DNA sequencing or other short-read DNA sequencing takes about or more than 10,000 genomes during library construction. Considering the seqeuncing depth is much lower than 10,000X, de novo insertions which only occurs in one or few genomes are only supported by one read. Therefore, TEMP2 regard singleton reads as potential de novo insertions by default. However, if a special library with limited number of genomes (for example: 50) is used during library constrcution, de novo insertions can be supported by more than one read, known as 2p or 1p1 insertions. In this situation, users can add a self-defined supporting-read cutoff using option "-C cutoff" to determine if a insertion is potential de novo insertion. For example, if 50 genomes are used in library construction and the seqeuncing depth is 50X (100bp paired reads, 400bp fragmnent length), the number of supporting reads for a de novo insertion should be around (fragment_length/read_length/2depth2/number_of_genome=400/100/2502/50=4), and you can set -C 7 for a more accurate de novo insertion estimation.
TEMP2-v0.1.1: Original release of TEMP2.