Halfpipe is designed for the estimation of RNA half-lives from time course data of RNA labeling protocols such as SLAM-seq or TimeLapse-seq. The pipeline builds on the modeling framework presented in Müller et al. (2024). To this end, it automates essential pre-processing, model fitting and post-processing steps. Please read the accompanying manuscript Müller et al. (2025) for more information about the underlying methodology of Halfpipe.
Required:
Optional:
- NVIDIA CUDA GPU(s)
Simply download this repository to a location of your choice. We recommend using a high performance machine for computation. A sample script slurm_example.sh that you can customize to run Halfpipe with the widely used SLURM workload manager is located in the docs/ folder.
Next, set up a conda environment using the halfpipe_condaenv.yml file in the config/ folder:
user@foo:~$ conda env create --file=config/halfpipe_condaenv.yml This yaml-file contains all modules that are necessary to execute Halfpipe. By default, the environment is called Halfpipe, but of course you can call it whatever you want. By default, you can activate the environment with:
user@foo:~$ conda activate Halfpipe In general Halfpipe is invoked with the halfpipe.py script:
user@foo:~$ python3 src/halfpipe.py subcommand A detailed description of the available sub-commands can be found below. But first we want to emphasize the use of the config.yml file. This file contains all the parameters that the user needs to specify, and here is a snippet to show
what this file looks like:
---
input:
samples: path/to/inputsamplesheet.tsv
bed: path/to/3UTR.bed
refgenome: path/to/genome.fa
snps: path/to/SNPfile.vcf
summaryname: yoursamplename
output: path/to/outputdirectory/ # output directory
...In general, we advise to only work with absolute instead of relative paths!
The parameter samples defines the path to the sample-sheet (TSV format required!) which lists the corresponding fastq(.gz)-files. You can view examplary sample sheets in the docs folder, namely docs/mocksamplesheet_singleend_onecompartment.tsv, docs/mocksamplesheet_singleend_twocompartment.tsv, docs/mocksamplesheet_pairedend_onecompartment.tsv, and docs/mocksamplesheet_pairedend_twocompartment.tsv. Column 1 (and 2, in the case of paired-end data) defines the path to the fastq(.gz) samples, while the columns define the measurement time in minutes, a descriptive sample name for readability, and information about the respective compartment of the two-compartment model (either "nucleus" or "cytosol", specify nuclear samples first).
The parameter bed specifies the path to the BED file containing genomically coordinated, annotated 3'UTRs. Such a file can be downloaded from the UCSC Table Browser.
The parameter refgenome defines the path to the fasta file of the reference genome. Please use the RefSeq genome fasta for GRCh38.
The parameter snps defines the path to the SNP annotation file in VCF format. You can download this file from the NCBI SNPdb database.
The parameter summaryname can be used to specify any name for the resulting summary file.
The parameter output defines the output path. All subfolders and files generated by halfpipe are saved under this path.
Further parameters such as the library type and the desired model can be configured in the config.yml file.
The pipe subcommand executes the entire pipeline for a 4sU time-series experiment. It maps (using NextGenMap) and filters reads from Fastq raw files, corrects the data for potential SNPs and RNA editing sites, and counts the T>C conversion induced by 4sU labeling. Halfpipe uses these counts to estimate the proportions of newly synthesized reads over time (EM algorithm) before fitting either a one- or two-compartment model of RNA metabolism to these quantities.
Usage:
user@foo:~$ python3 halfpipe.py pipe The mapandfilter subcommand maps and filters reads from raw fastq-files using NextGenMap. The output BAM-files can be found in the mapped/ and filtered/ subfolders.
user@foo:~$ python3 halfpipe.py mapandfilter The subcommand readpreprocess is used to identify putative editing sites (including SNPs, A-to-I editing sites, etc.) from control samples. This step is crucial for the later calculation of the labeling efficiency with the ratioestimation subcommand. If this step is not performed, false-positive T>C conversions would bias the downstream analyses. The output files can be found in the readpreprocess/ subfolder.
user@foo:~$ python3 halfpipe.py readpreprocess The subcommand ratioestimation is used to estimate the 4sU incorporation rate and accordingly the ratio of new to total RNA for each measured 3'UTR. These quantities are the targets for model fitting. The output files can be found in the subfolder ratioestimation/.
Usage:
user@foo:~$ python3 halfpipe.py ratioestimation The subcommand createsummary summarizes the individual results of the ratioestimation. The resulting output file is required for model fitting and can be found in the subfolder summaryfiles/.
Usage:
user@foo:~$ python3 halfpipe.py createsummary The subcommand fitparameters applies either the one- or two-compartment model to estimate the RNA half-lives. The output files can be found in the parameterfit/ subfolder.
Usage:
user@foo:~$ python3 halfpipe.py fitparameters