DQRbuild Toolkit Investigations of significant alerts distributed by LIGO-Virgo in O3

This dataset accompanies the paper "Rapid data quality investigations of gravitational-wave events with the Data Quality Report Builder toolkit" (Davis et al. 2026, arXiv:2605.16183). It contains the results of manual data quality vetting and automated statistical tests applied to all significant gravitational-wave candidates shared as public low-latency alerts during the third LIGO-Virgo observing run (O3, April 2019 – March 2020). --- Contents updated_dqr_replay.csv Results of the manual data quality review (DQR) for 79 O3 low-latency alert candidates, corresponding to Tables 1 and 2 in the paper. Each row represents one superevent. Columns: Event: GraceDB superevent identifier (e.g., S190408an) DQR Status: submission status of the DQR EV Conclusion: free-text summary of the expert vetter's findings, including notes on specific tool outputs (stationarity, glitchfind, omega scan, hveto, pemcheck, gspynettree, etc.) and an overall assessment O3 conclusion: final O3 outcome label ("Retract" or "DQ issue"); blank if no data quality concern was identified H1 result: LIGO Hanford detector-level conclusion ("DQ issue", "-" for not applicable, or blank) L1 result: LIGO Livingston detector-level conclusion V1 result: Virgo detector-level conclusion Of the 79 candidates, 23 were ultimately retracted, and 10 were flagged as having data quality issues. --- O3_DQR_test_results.hdf5 Results of the automated DQRbuild statistical tests run on the same 79 O3 candidates. The file is stored in HDF5 format and is readable with standard scientific Python libraries (e.g., h5py, gwpy.table.EventTable). The top-level structure contains: superevent_id: GraceDB superevent identifier, matching the Event column in the CSV ifo_state/H1, ifo_state/L1, ifo_state/V1: Boolean flags indicating whether each interferometer was observing at event time Per-detector test result groups (H1/, L1/, V1/), each containing numerical outputs for the following checks: omegascan, stationarity, glitchfind, rangecheck, gspynettree, lockcheck, omega_overlap, hveto, pemcheck, and glitchaverage Network/gps_numerology: network-level timing check result --- Usage notes The superevent_id field in the HDF5 file matches the Event column in the CSV and can be used to join the two files. Not all detectors were observing for every event; fields for non-participating detectors are marked with a dash in the CSV (a blank field indicates no DQ issue was identified). The EV Conclusion column contains informal notes recorded during rapid vetting. The HDF5 file conforms to the __astropy_table__ metadata convention and can be read directly with astropy.table.Table.read(). --- Related resources GraceDB public alert archive: https://gracedb.ligo.org Raw data access: https://gwosc.org