Data for the GeneFamHunter sliding-window benchmarks: genome assemblies, GeneFamHunter search outputs and the NCBI CDD_delta database

Description This record holds the large inputs for the GeneHunter_benchmarks repository. That repository reproduces the sliding-window benchmarks of GeneFamHunter, a pipeline for finding the members of a gene family in genome assemblies (Klimov and He, Bioinformatics, submitted). Before searching, GeneFamHunter cuts each genome into overlapping 1,400-bp windows and translates them in six frames. The benchmarks ask whether this windowing changes what the homology searches find. Each search on windows (HMMER, RPS-BLAST, PSI-BLAST, DELTA-BLAST, and TBLASTN for comparison) is compared with the same search on whole contigs, on open reading frames, and on 290-kb windows. The test family is the papain-like cysteine proteases (Peptidase C1, Pfam PF00112). There are two datasets: qfo: six genomes from the Quest for Orthologs (QfO) reference proteomes, release 2026_02 (C. elegans, D. melanogaster, T. castaneum, D. magna, D. discoideum, P. falciparum). mites_shi2023: six astigmatid and psoroptid mite genomes analysed by Shi et al. (2023). Files genefamhunter_runs.tar (286 MB): for each of the 12 genomes, the folder genefamhunter_runs/<dataset>/<genome>/ contains: the genome assembly (genome.fasta.gz) GeneFamHunter’s search outputs on the windows: hmmer.domtblout, hmmer.domtblout.tab, rpstblastn.out2, psi_blast.out, delta_blast.out2 the regions of interest it identified: total_evidence.merged, plus total_evidence.merged2 for the mites cdd_delta.tar (3.1 GB): the NCBI CDD_delta database used by DELTA-BLAST, November 2019 version. NCBI distributes only the current version. The windows themselves (13 GB) are not included. The repository’s scripts/prepare_windows.sh regenerates them exactly and checks them against MD5 sums. Usage Clone the GeneHunter_benchmarks repository, then run bash scripts/get_data.sh. It downloads both archives, unpacks them into data_large/ and checks every file against data/checksums/data_large.md5. Then run bash scripts/run_all.sh qfo and bash scripts/run_all.sh mites_shi2023. Each run ends by comparing its tables with the published results in expected/. The tool versions are listed in environment.yml (HMMER 3.3.2, BLAST+ 2.13.0). Sources and licences Data derived in this work (GeneFamHunter outputs and regions of interest): CC BY 4.0. Third-party data, redistributed unchanged under their own terms: QfO genome assemblies from INSDC (GenBank/ENA/DDBJ); accessions are in data/genome_lists/qfo.tsv. Mite genome assemblies received from the authors of Shi et al. (2023), Clin. Transl. Allergy 13, e12324 (doi:10.1002/clt2.12324), and redistributed with their permission; GenBank accessions are in data/genome_lists/mites_shi2023.tsv. If you use these genomes, cite Shi et al. (2023). NCBI CDD_delta: public domain. Details: see LICENSE-DATA.md in the repository.