Flawed methods in “COVID-19: Attacks the 1-Beta Chain of Hemoglobin and Captures the Porphyrin to Inhibit Human Heme Metabolism”

This is a critical commentary on an earlier submission by Liu and Li. The preprint from Liu & Li (https://doi.org/10.26434/chemrxiv.11938173.v7) puts forward hypotheses about a proposed role for proteins of SARS-CoV-2, the virus associated with Covid-19, in directly attacking haemoglobin in patients’ blood. Arguments for the hypotheses are based on computational methods: bioinformatics calculations searching for evidence that viral proteins share functional domains related to haem binding with human proteins, molecular modeling of viral proteins, and computational docking of these protein models with models of haem, porphyrin and haemoglobin. No experimental evidence is provided to support any of the conclusions. When interpreted according to accepted standards, these computational results do not hold up and do not provide support for the hypotheses. The interpretation of the search for shared functional domains suffers from a fundamental error in how the significance of the results is judged; when interpreted correctly, there is no evidence for these shared functional domains. Molecular modeling is carried out with tools that are easy to use but not best-in-class, and no allowance is made for uncertainty in the resulting atomic coordinates. Finally, the docking results are invalidated by a catastrophic error in their interpretation: the authors choose the docking trials that have the highest energies, whereas the most stable complexes are actually the ones that have the lowest energies and are therefore least strained. An addendum addresses flaws in a new version 8 from Liu & Li (https://doi.org/10.26434/chemrxiv.11938173.v8), which retracts most of their results from earlier versions but nonetheless continues to put forward the same conclusions on the basis of poorly-controlled docking calculations.