Data for "A replicable and modular benchmark for long-read transcript quantification methods"

This dataset was used in the paper titled "A replicable and modular benchmark for long-read transcript quantification methods". It contains simulated ONT and PacBio datasets as follows:

Mouse PacBio and ONT Datasets:

The PacBio dataset is available at Zenodo record 11201284. It was simulated with an overall error rate of 1.4%, following the methodology described in Prjibelski et al., 2023.

The ONT dataset was derived from the R10.4 dataset available at Zenodo record 7121404. Reads were extracted by converting the BAM file to FASTQ format and then shuffled using the shuffle.sh utility from the BBMap suite. Simulation was performed using NanoSim with an error rate of 2.8%. The default NanoSim behavior of introducing background "unaligned" reads was disabled to ensure a cleaner simulation.

ONT cDNA and Direct RNA Datasets (NA12878 cell line):

These datasets were simulated using NanoSim based on the NA12878 sample from the Nanopore WGS Consortium.

ONT Direct RNA Dataset with Realistic Truncation (NA12878 cell line): 

This dataset was simulated using a combination of DRSim and Badread, based on the NA12878 sample. DRSim was used to model realistic 3′-end truncation patterns, while Badread simulated the Nanopore sequencing error profile. Further details are provided in the paper referenced above.

This archive contains all input files necessary to run the initial benchmarks associated with the lr_quant_benchmarks repository.

The archive can be decompressed with tar and zstd using the command tar --use-compress-program=zstd -xf input.tar.zstd