Phasertng: directed acyclic graphs for crystallographic phasing

Abstract

Crystallographic phasing strategies increasingly require the exploration and ranking of many hypotheses about the number, types and positions of atoms, molecules and/or molecular fragments in the unit cell, each with only a small chance of being correct. Accelerating this move has been improvements in phasing methods, which are now able to extract phase information from the placement of very small fragments of structure, from weak experimental phasing signal or from combinations of molecular replacement and experimental phasing information. Describing phasing in terms of a directed acyclic graph allows graph-management software to track and manage the path to structure solution. The crystallographic software supporting the graph data structure must be strictly modular so that nodes in the graph are efficiently generated by the encapsulated functionality. To this end, the development of new software, Phasertng, which uses directed acyclic graphs natively for input/output, has been initiated. In Phasertng, the codebase of Phaser has been rebuilt, with an emphasis on modularity, on scripting, on speed and on continuing algorithm development. As a first application of phasertng, its advantages are demonstrated in the context of phasertng.xtricorder, a tool to analyse and triage merged data in preparation for molecular replacement or experimental phasing. The description of the phasing strategy with directed acyclic graphs is a generalization that extends beyond the functionality of Phasertng, as it can incorporate results from bioinformatics and other crystallographic tools, and will facilitate multifaceted search strategies, dynamic ranking of alternative search pathways and the exploitation of machine learning to further improve phasing strategies.

Keywords: Phaser; Phasertng; SAD phasing; directed acyclic graphs; machine learning; maximum likelihood; molecular replacement.

Figure 1

(a) Path, (b) tree, (c) directed acyclic graph. The root node is shown in red, leaf nodes are shown in green and intermediate are shown in blue, except that all nodes with two parents are shown in purple.

Figure 2

(a) Schematic for the DAG resulting from user input and phasertng.xtricorder with a colouring scheme as in Fig. 1 ▸. Nodes outlined in orange are those generated by phasertng.xtricorder. Multiple arrows indicate where the DAG may branch. (b) Schematic of DAG for structure solution of PDB entry 4n3e (Sliwiak et al., 2014 ▸). The crystal was obtained by co-crystallization of Hyp-1 with an eightfold molar excess of the ligand 8-anilino-1-naphthalene sulfonate (ANS). Strong fluorescence under UV illumination confirmed the presence of ANS in the crystals. Data were collected from a single crystal at a wavelength of 1.00 Å. The data extended to 2.4 Å resolution. There were no systematic absences of reflections along the axes and the highest symmetry in which the data merged was space group P422. Significant anisotropy was present. TNCS of order 7 was suspected as a result of analysis of the Patterson map, with the absence of TNCS maintained as a hypothesis. Twinning was detected in the intensity statistics after TNCS corrections for order 7. Since overmerging is possible in the presence of twinning, data were expanded to all subgroups of P422. Twinning was not detected in the absence of TNCS. At the conclusion of phasertng.xtricorder data analysis there are eight hypotheses for the contents of the asymmetric unit; solutions can be obtained for two of these hypotheses.

Figure 3

Number of reflections in each test case for the database of 30 test cases used for the speed tests in Figs. 4 ▸ and 5 ▸. PDB identifiers are shown for data with no TNCS (orange), for data with TNCS of order 2 (light blue) and data with TNCS of order greater than 2 (dark blue).

Figure 4

Comparison of the fold speedup of wall time for Phaser and Phasertng with and without threading over five threads. (a) Data with no TNCS, (b) data with TNCS of order 2 and (c) data with TNCS of order greater than 2. Four times are shown for each PDB identifier: Phaser without threading (blue), Phaser threaded on five cores (red), Phasertng without threading (green) and Phasertng threaded on five cores (purple). The longest runtime for each PDB identifier is shown in seconds above each column group, which is for Phaser without threading in every case.

Figure 5

Comparison of the fold speedup of wall time for Phaser (red) and Phasertng (purple) with threading for between one and five threads. The elapsed wall time is shown above each column in seconds for (a) data with no TNCS (PDB entry 1za7), (b) data with TNCS of order 2 (PDB entry 1hto) and (c) data with TNCS of order greater than 2 (PDB entry 4n3e).