Currently performs transcript identification, gene assignment, and naive quantification. A more sophisticated quantification method remain on the TO-DO list.
IMPAQT (Identifies Multiple Peaks and Quantifies Transcripts) is a transcript identification and gene expression quantification method for TAGseq and 3' mRNAseq experiments. It operates on assumptions about the distribution of reads along the 3' UTR of expressed genes. Clustering these reads enables pseudo-transcript identification and quantification of expression at the gene and transcript level for isoforms utilizing distinct 3' ends.
It generates a GTF file defining the boundaries of each transcript and their expression level as well as, optionally, a gene expression counts table if a reference annotation is provided.
This method is particularly useful in non-model organisms where 3' UTRs for most genes are poorly annotated (resulting in massive data loss). Increased gene density also tends to hurt the assignment of transcripts by this aglorithm, as it increases assignment ambiguity. Reads for which a reasonable transcript of origin cannot be identified are handled individually.
- Make sure cmake and make are installed on your machine.
# Linux
sudo apt install cmake zlib1g-dev
# Mac
brew install cmake zlib
- Clone this repository and change into it.
git clone https://github.com/bnjenner/impaqt.git
cd impaqt
- Create a build directory and change into it.
mkdir build
cd build
- Compile
cmake ../
make
- Install
sudo make install
- Give it a go!
impaqt input.sorted.bam
SYNOPSIS
impaqt input.sorted.bam [options]
DESCRIPTION
Identifies Multiple Peaks and Qauntifies Transcripts. Identifies and quantifies isoforms utilizing distinct 3'
ends. Generates a GTF file of identified transcripts and optionally a counts file written to stdout if a reference
annotation is provided.
REQUIRED ARGUMENTS
BAM INPUT_FILE
OPTIONS
-h, --help
Display the help message.
-t, --threads INTEGER
Number of processers for multithreading. Default: 1.
-a, --annotation INPUT_FILE
Annotation File (GTF or GFF). If specified, a counts table will be output through standard out. NOTICE: File
type identified by file extension. Default: .
-s, --strandedness STRING
Strandedness of library. One of forward and reverse. Default: forward.
-n, --nonunique-alignments
Count primary and secondary read alignments.
-q, --mapq-min INTEGER
Minimum mapping quality score to consider. Default: 1.
-w, --window-size INTEGER
Window size to use to parition genome for read collection. Default: 1000.
-m, --min-count INTEGER
Minimum read count to initiate DBSCAN transcript identification algorithm. (Hard minimum of 10) Default: 25.
-p, --count-percentage INTEGER
Minimum read count percentage for identifying core reads in DBSCAN algorithm. This will be the threshold
unless number of reads is less than 10. Default: 5.
-e, --epsilon INTEGER
Distance (in base pairs) for neighboring reads in DBSCAN algorithm. This should generally be 0.5-1.5x the
read length, depending on desired isoform sensitivity (lower = more sensitive). Default: 50.
-d, --density-threshold DOUBLE
Read density threshold (# reads / # bps) to skip transcript identification. Assignment in super dense
regions (usually the mitochrondria) doesn't really benefit from transcript identificaiton. Default is unset.
Default: 0.
-f, --feature-tag STRING
Name of feature in GTF for assignment. Default: exon.
-u, --utr-tag STRING
Name of UTR feature in GTF for assignment. Default: UTR.
-i, --feature-id STRING
ID of feature to use for feature assignment. Default: gene_id.
-o, --output-gtf STRING
Specify name of cluster GTF file. Default is BAM name + ".gtf".
--version
Display version information.
VERSION
Last update: August 2025
impaqt version: beta
SeqAn version: 2.4.0
Utilizes libraries like bamtools and seqan and the DBSCAN algorithm is inspired by EmbeddedArtistry and github user Eleobert.
For questions or comments, please contact Bradley Jenner at bnjenner@bu.edu