Greetings — I am interested to know more about the following points to understand the results :
[1] How was the “C-alpha-IDDT” (Mariani et. al., Bioinformatics, 29(21), 2722-2728, 2013) used, as - if I understand, the unprecedented and exceptional prediction capabilities of AlphaFold2 - as compared with the second place team’s program (which program was this?) - rests entirely on how this score is obtained — or, how is the GDT score related to the C-alpha-IDDT. (Is the GDT score calculation published?) Is the C-alpha-IDDT score really based solely on C-alpha? Does CASP from 2006 through the current CASP (CASP14?) use the same score, and is it the C-alpha-IDDT? Because the C-alpha-IDDT score was published in 2013, and the top scores - as seen in the histogram in Nature News are roughly flat until 2014, after which the top score appears to increase linearly. Could the programs use this score to select solutions, or given that the programs can be using difference distance matrices anyway, isn’t it expected to produce a higher score than the e.g. RMSD on the peptide backbone? [2] How exactly are the “poor” predictions of structures determined by NMR to be explained, given the excellent predictions otherwise, and how does the overall score in the Nature News histogram account for this discrepancy? [3] how did the training data set increase in size, particularly from 2014? More cryo-EM structures? [4] what “additional information about the physical and geometric constraints that determine how a protein folds” was used in AlphaFold2, when was it discovered, did other teams use this “additional information”, and is this the first year it was used? Thanks, -Bryan References : Nature News column histogram: https://media.nature.com/lw800/magazine-assets/d41586-020-03348-4/d41586-020-03348-4_18633154.jpg CASP book of abstracts: https://predictioncenter.org/casp14/doc/CASP14_Abstracts.pdf (year 2020, I assume) ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
