State-of-the-Art Crystallographic Refereeing General Message: Deposition of Reflection Data should be Obligitary Why? Three examples Ton Spek, Utrecht University, The Netherlands ACA-Honolulu - July 23, 2013 The Purpose of Refereeing an X-RAY Study is to Check: • Is the supplied information complete for a meaningful structure report and evaluation? • Is the data quality adequate for the conclusions drawn? • Are proper data collection and structure analysis techniques used? • Is the structure correct? • Is the interpretation of the result correct? • Are there additional observations of interest but missed by the authors? • Is this a duplication with no added value or even a fraud? Data Needed for Proper Refereeing • The Refinement Model (Coordinates, con/restraints etc.). • The reflection data on which the Model is based. • For Acta Cryst. this implies the deposition of a parameter ‘CIF’ and an Fobs/Fcalc ‘FCF’. • The new SHELXL2013 offers an option for the inclusion of both types of information in a single (CIF) file. • This allows for the creation of a proper automated validation report and data archival. • It also provides an option for additional independent calculations where relevant. IUCr CheckCIF Validation • The IUCr has set up an elaborate validation system for structures published in its own Acta Cryst. Journals. • Most chemical journals now also require authors to deposit a CIF and to supply a validation report for use by the referees but do not require the deposition of the reflection data. • The problem is that authors and referees still have to understand the implication of ALERT issues. • Even with Acta Cryst. the above sometimes fails as shown in the next example (Paper withdrawn now). Acta Cryst. (2011). E67, m576-m577 Is this structure correct ? [Co4Cl4(C3H6S)4] R1 = 0.041 wR2 = 0.125 Co Zn C3 N1 C6 N2 R1 = 0.028 wR2 = 0.072 Identical to: Inorg. Chem. (2006) 45, 8318 Correction proved with reflection data Problem Example from Chemical Journal Biswas et al. Eur. J. Inorg. Chem. (2012) 4479-4485. R1 = 0.08, wR2 = 0.30 Space Group C2/c Fe-Complex on 2-axis NO3 Anion in general pos. Population = 0.5 !? Problematic Structure (Biswas et al.) • Problem: NO3- in general position disordered over two symmetry related sites sounds strange (precedents ?). • Significant residual density peaks up to 2.5 e/Å3 but explained away as ‘With no chemical significance’ • Reflection data would be needed for a detailed analysis of the issue but only a CIF was deposited. • Martin Lutz et al. resynthesised the compound following the Biswas et al. protocol with a surprising result (Eur. J. Inorg. Chem. (2013) 2467-2469) • C2/c Cc and NO3 NO3 + Acetic acid + H2O Biswas et al. Lutz et al. C2/c – Biswas et al. - VOIDS Cc – Lutz et al. Follow-Up - Biswas et al. maintain their rebuttal [Eur. J. Inorg. Chem. (2013) 2470] with their interpretation. - Referee conclusion (with no access to the reflection data): Two different structures !!! - I obtained the reflection data from one of the co-authors - A SQUEEZE test run on the Biswas et al. data, with the NO3- removed, recovered 279 electrons from the voids left where only 128 were expected for an NO3- model. - Run on Lutz et al. data, with NO3 and solvents removed, recovered 285 electrons of the expected 296. Conclusion ? Widely Cited Nature paper in the News Nature (2013) 495, 461 Did a Crystallographic Referee have a serious look at the data ? • The deposited CIFs bring up multiple validation issues in need of further and detailed analysis. • Important information is lacking to justify the use of the SQUEEZE technique in this MOF context. Its use is probably not valid in this application. • Of particular interest are the details of the reported structure determination of miyakosyne A with a population of 50%. • No reflection data have been deposited to allow for an independent analysis of the structures. From Nature Paper Major Important Error Types • • • • • • Wrong atom type assignments (e.g. Cu for Zn) Problems with hydrogen atoms (too few or too many etc.) Missed twinning Missed disordered solvent Artificial ‘disorder’ due to wrong structure analysis Wrong symmetry • All the above may lead to one strong ALERT but more often it leads to multiple weaker ALERTS that in combination would indicate a strong ALERT. • We usully need the reflection data to sort out those ALERTS Plea for the deposition of both the ‘CIF’ and the reflection data • A minimum would be the parameter CIF and the Fo/Fc data from the final refinement (FCF). • Better: Embed the refinement details and the observed data in the CIF. The IUCr/Acta Cryst. provides for that purpose the data names _iucr_refine_instructions_details and _iucr_refine_reflections_details • The new SHELXL2013 does this by embedding automatically (with data names starting with _shelx_) and adds checksums. Final Comments • CheckCIF provides a list of issues that need to be addressed, either with a correction or giving an acceptable explanation. • ALERTS come in levels A, B & C and report possible error, missing data or quality issues. G-ALERTS are generally informative and not necessarily errors but still need to be looked at seriously. • A & B ALERTS should never automatically lead to the rejection of a paper but should be investigated by a professional crystallographer. • Everything unusual marked as ‘new’, ‘unexpected’ etc. in a paper should be approached with healthy mistrust. For the Discussion • Do we need different validation criteria for papers in crystallographic journals such as Acta Cryst. and papers published in chemical journals ? • Should deposition of the reflection data be required by all journals (Standard now in the macromolecular crystallography community) ? • Should the validation report along with (where relevant) the authors’ comments be published as supplementary material ? Thanks to George Sheldrick for the major 2013 update of SHELXL and my former coworker Martin Lutz and many others for useful comments and suggestions