Skip to main content

Splice aligner of long transcriptomic reads to genome.

Project description


uLTRA is a tool for splice alignment of long transcriptomic reads to a genome, guided by a database of exon annotations. uLTRA takes reads in fast(a/q) and a genome annotation as input and outputs a SAM-file. The SAM-file includes information on which splice sites are found and if the read is a full splice match (and to which transcript), incomplete splice match, Novel in catalog, or novel not in the catalog, as defined in SQANTI. uLTRA is particularly accurate when aligning to small exons see some examples.

uLTRA is distributed as a python package supported on Linux / OSX with python v>=3.4. Build Status.

Table of Contents


There is a bioconda recipe, docker image, and a singularity container of uLTRA v0.0.3.3 created by sguizard. You can use, e.g., the bioconda recipe for an easy automated installation.

If a newer version of uLTRA is not available through bioconda (or you simply want more control of/customize your installation), alternative ways of installations are provided below. Current version of uLTRA is 0.0.4 (see changelog at end of this readme).

Using conda

Conda is the preferred way to install uLTRA. You can either clone this repository and run the script or you can perform step 1-6 below manually for more control.

Installation with script

git clone --depth 1
cd uLTRA
./ [An install directory in your PATH]

The install script is tested in bash environment and will perform the steps 1-6 below automatically for you.

You need to activate the conda environment "ultra" to run uLTRA as:

conda activate ultra

Manual installation

1. Create and activate a new environment called ultra

conda create -n ultra python=3 pip 
conda activate ultra

2. Install uLTRA

pip install ultra-bioinformatics

3. Install third party MEM finder slaMEM and aligner minimap2

git clone
cd slaMEM

Place the generated binary slaMEM in your path. Minimap2 can be installed through conda with conda install -c bioconda minimap2, or manually.

4. You should now have 'uLTRA' installed; try it:

uLTRA --help

Upon start/login to your server/computer you need to activate the conda environment "ultra" to run uLTRA as:

conda activate ultra

5. Test uLTRA

Download/use test data available in this repository here and run:

uLTRA pipeline [/your/full/path/to/test]/SIRV_genes.fasta  \
               /your/full/path/to/test/SIRV_genes_C_170612a.gtf  \
               [/your/full/path/to/test]/reads.fa outfolder/  [optional parameters]

Specify the absolute path to the GTF-file on your system, otherwise gffutils will complain and giva a cryptic ValueError: unknown url type: error message. Outfile will be outfolder/reads.sam, unless you specify your custom prefix filename with --prefix.

6.(Optional) Install of StrobeMap

Using NAM seeds is new since version 0.0.4. It can reduce runtime, disk usage and provide fixe memory usage to default MEM findin. See changlog at end of this file. StrobeMap is installed on Linux with:

mv StrobeMap-0.0.2 StrobeMap
chmod +x StrobeMap

Place the generated binary StrobeMap in your path.

7. (Optional) Install of MUMmer

While MUMmer is usually not used in uLTRA, if slaMEM fails, uLTRA falls back on finding MEMs with MUMmer until the slaMEM bug has been fixed. In this corner case, uLTRA needs MUMmer avaialble in the path. MUMmer can be installed with

conda install --yes -c bioconda mummer

Downloading source from GitHub

Make sure the below-listed dependencies are installed (installation links below). Versions in parenthesis are suggested as uLTRA has not been tested with earlier versions of these libraries. However, uLTRA may also work with earlier versions of these libraries. All below dependencies except slaMEM can be installed as pip install X or through conda.

With these dependencies installed. Run

git clone
cd uLTRA


uLTRA can be used with either PacBio Iso-Seq or ONT cDNA/dRNA reads.


First, we construct the data structures used in uLTRA using a genome annotation GTF file and a genome fasta file. Make sure to specify full path to annotation, otherwise gffutils will complain.

uLTRA index genome.fasta  /full/path/to/annotation.gtf  outfolder/  [parameters]


For example

uLTRA align genome.fasta reads.[fa/fq] outfolder/  --ont --t 8   # ONT cDNA reads using 8 cores
uLTRA align genome.fasta reads.[fa/fq] outfolder/  --isoseq --t 8 # PacBio isoseq reads

You can set a custom location of where to get the index from using --index [PATH]. Otherwise, uLTRA will try to read the index from the outfolder/ by default. The aligned reads will be written to outfolder/reads.sam unless --prefix is set. For example, --prefix sample_X will output the reads in outfolder/sample_X.sam.


Performs all the steps in one

uLTRA pipeline genome.fasta /full/path/to/annotation.gtf reads.fa outfolder/  [parameters]


uLTRA outputs a SAM-file with alignments to the genome. In addition, it outputs to extra tags describing whether all the splices sites are known and annotated (FSM), new splice combinations (NIC), etc. For details see the definitions of notations in the Sqanti paper.


Please cite [1] when using uLTRA. If you are using uLTRA v0.0.2 or later please also cite minimap2 as uLTRA incorporates minimap2 for alignment of some reads. For example "We aligned reads to the genome using uLTRA [1], which incorporates minimap2 [CIT].".

  1. Kristoffer Sahlin, Veli Makinen. 2020. "Accurate spliced alignment of long RNA sequencing reads" paper available here.

Bib record:

@article{10.1093/bioinformatics/btab540, author = {Sahlin, Kristoffer and Mäkinen, Veli}, title = "{Accurate spliced alignment of long RNA sequencing reads}", journal = {Bioinformatics}, year = {2021}, month = {07}, abstract = "{Long-read RNA sequencing technologies are establishing themselves as the primary techniques to detect novel isoforms, and many such analyses are dependent on read alignments. However, the error rate and sequencing length of the reads create new challenges for accurately aligning them, particularly around small exons.We present an alignment method uLTRA for long RNA sequencing reads based on a novel two-pass collinear chaining algorithm. We show that uLTRA produces higher accuracy over state-of-the-art aligners with substantially higher accuracy for small exons on simulated and synthetic data. On simulated data, uLTRA achieves an accuracy of about 60\% for exons of length 10 nucleotides or smaller and close to 90\% accuracy for exons of length between 11 to 20 nucleotides. On biological data where true read location is unknown, we show several examples where uLTRA aligns to known and novel isoforms containing small exons that are not detected with other aligners. While uLTRA obtains its accuracy using annotations, it can also be used as a wrapper around minimap2 to align reads outside annotated regions.uLTRA is available at data are available at Bioinformatics online.}", issn = {1367-4803}, doi = {10.1093/bioinformatics/btab540}, url = {}, note = {btab540}, eprint = {}, }


GPL v3.0, see LICENSE.txt.


New since v0.0.4

An option --use_NAM_seeds is added. This parameter changes the seeding of MEMs to NAMs (with strobemers). If --use_NAM_seeds is specified, uLTRA calls StrobeMap (binary can be acquired here and put StrobeMap in your path). This seeding results in uLTRA being faster (say 15-70% depending on the number of threads) and produces much smaller intermediate files. However, the memory usage is fixed ~80-90Gb for human genome (regardless of number of cores/threads). This makes --use_NAM_seeds less memory consuming than the default option (MEMs) when uLTRA is given about 18 cores or more (with --t), and more memory consuming than the default version for t < 18. The alignment accuracy is largely the same. Using NAM seeds decreases the accuracy of about 0.01%-0.05% compared to MEMs (i.e., 1 alignment in every 2,000-10,000). Accuracy was measured as all correct splice sites identified compared to at least one nt alignment offset from a correct splice site.

Due to the "fixed" memory usage, faster runtime, and smaller intermediate file size, I recommend --use_NAM_seeds option for large datasets (>5M reads) if running on nodes with more than 20 cores.

New since v0.0.3

uLTRA now uses less than half of the memory used in previous versions and is about 20% faster.

New since v0.0.2

Since v0.0.2, uLTRA can be used as an end-to-end aligner for annotation and detection of novel genes or isoforms (default mode). This is because uLTRA (>=v0.0.2) now incorporates minimap2. minimap2 is run upon start of uLTRA, and the results are used both for (i) not aligning reads with uLTRA which had a primary alignment to regions not indexed by uLTRA (e.g. genomic regions or unannotated genes) and (ii) to consult at end of program which aligner had a better fit (based on cigar) of the primary alignment and chose this alignment to be primary. uLTRA still uses its own alignment algorithm to align to and around all annotated gene regions. uLTRA can therefore, at worst, be seen as an advanced wrapper around minimap2 that refines alignments around annotated regions. See updated CREDITS when using this version. uLTRA can still be used as a stand alone aligner as presented in our paper by specifying --disable_mm2.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for ultra-bioinformatics, version 0.0.4
Filename, size File type Python version Upload date Hashes
Filename, size ultra_bioinformatics-0.0.4.tar.gz (58.8 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page