Speaker
Description
Automated data collection is a central component of modern macromolecular crystallography beamlines, enabling rapid and reproducible acquisition of diffraction data with minimal user intervention. A typical automated workflow integrates robotic sample mounting, optical pre-centering, X-ray-based centering, strategy calculation, and diffraction data acquisition.
At XALOC, one of the two MX beamlines at the ALBA synchrotron, we have implemented unattended, fully automated data collection in the PyQt version of MXCuBE, the data-acquisition software. After initial coarse and fine optical centering, the crystal position is optimized using an X-ray centering procedure designed to locate regions of best diffraction, using DOZOR [1] for evaluation of the quality of the diffraction pattern, eliminating powder and ice rings. This procedure uses a two-dimensional (2D) raster scan across the sample to identify candidate diffracting regions. The region with the highest DOZOR score is then refined by a one-dimensional (1D) line scan performed after a 90° rotation of the sample, providing excellent sample alignment along the beam direction.
The procedure focuses on maximizing success rate and ensuring reliable sample detection. In addition, automated data collection (ADC) has significantly improved both speed and reliability. Sample mounting and automated centering can be achieved within 3-5 minutes per sample and data collection is then performed at 100 images/sec. The success rate of the ADC exceeds 98%, approaching 100%. Cases in which crystals cannot be located are mainly related to the trade-off between the size of the 2D meshes used in X-ray centering and the size and positioning of the sample loops; in some instances, crystals mounted in large loops close to the metal pin may fall outside the scanned region. Data quality is in most cases improved, or at least comparable to that obtained by manual centering. Users can follow the progression of the ADC in real time using the LIMS web page and download automatically generated processed data and density maps after molecular replacement.
The procedure implemented at XALOC enables high-throughput screening, reduces operator variability, and improves reproducibility. This is of particular advantage for Fragment-based drug discovery (FBDD) currently implemented at ALBA. Ongoing developments in computer vision, adaptive scanning, and machine-learning-based decision-making are expected to further enhance centering accuracy, throughput, and robustness in fully automated crystallographic pipelines.