Talk:Use case "RADDOSE"

From EdnaWiki
Jump to: navigation, search

General

  • Is RADDOSE protein Xtal specific or is it able to deal with pure polynucleotides Xtal as well?
    • In principle, it is not specific to any chemistry at all: you can do without the NRES/NRNA/NDNA keywords to input a complete chemical description of the unit cell contents, for example:
  PATM C 1000 N 270 O 300 H 1600 SE 4 S 10 ZN 4 P 12
  SATM  Na 1000 Cl 1000 ....
  SOLVENT 50%
  NMON 1

The parameters to PATM could be calculated from sequence(s) and information about the number of each chain in the unit cell (or asymmetric unit + spacegroup). This is not protein specific at all. In the RADDOSE documentation of the PATM keyword, the term "protein part" should be understood to mean "non-bulk-solvent part". Very confusing! --Peter

    • OK, I felt indeed that it was not necessary protein-specific, but when reading the RADDOSE documentation, especially the synopsis ("calculations of absorbed X-ray dose by **protein** crystals"), it did not sound very clear. -- Marie-Francoise


Input

  • Energy or Wavelength is an optional parameter in RADDOSE. Default value is related to the copper K-alpha radiation... shall we keep this default value for the prototype?
    • I suggest using a real wavelength: this is easily obtained from bestfile.par as written out by MOSFLM. Radiation damage is heavily dependent on wavelength, so the prototype will be a lot less realistic if we use one that is far too long. Also, if we leave this out, anyone looking at the prototype will home in on this omission straight away and it will make the prototype look very bad. --Peter.
    • I agree on the fact that radiation damage is wavelength dependent, that's why I've put a note whether it should be optional in the first version of this page. For now, I let whether we will take the wavelength or the energy as a question mark (to be discussed in a next VC?). Regarding how to retrieve this value, I think that the wavelength (or energy) should be known at the beginning of the experiment, so that this parameter should be passed directly from the "global/inital" dataset rather than passing it from the MOSFLM plugin to the RADDOSE one (especially if we decide not to use MOSFLM but another software for the integration). -- MF
    • wavelength or energy? a priori wavelength (most of the programs use wavelength ie BEST), but this shouldn't a big deal to use wavelength or energy by a simple conversion or by retrieving it from the diffraction plan (?). -- MF
  • Unit cell size: can be either unit cell parameter or volume. As MOSFLM outputs the unit cell parameters, I suggest choosing the unit cell parameters here. -- MF
  • Chemical composition:
    • The important parameter is the composition and amount of solvent in the crystal, if the solvent proportion isn't known, one needs to know the chemical composition of the crystal, NRES, NDNA, NRNA (see raddose user guide) what about NMON? We should determine the required/optional parameters for the chemical composition.
    • there are different means to define the chemical composition of a sample (CHON): NRES/NDNA/NRNA or PATM. Peter suggested that it would be possible to determine the PATM line given the compound code or sequence (with the condition that the code or sequence should be known by the user). There is an example python application for calculating PATM from sequences and known chemical components in the crystal, to illustrate my thinking about this. -- Peter.
      • Should we go that way or should we keep the possibility to define the chemical composition using both NRES/NDNA/NRNA and PATM?
      • Is there cases where the user knows only the number of residues so that NRES should be needed?
      • Some answers from Gleb: It is very elegant to retrive number of sulfurs for PATM and NRES from the sequence by code, thought there are many exceptions (Se, Ligands etc.). DNA and RNA sequences would rearly be possible to get from the database, though it is convenient for user to paste the sequence. Many users will be reluctant to input the sequence/code (disclosing their top-secret-crystallizable construct...). NRES/NRNA/NDNA , or especially the molecular weight of those (easy to covert to N... ) people normally know by hart. The cases when they only know N... are rear but do exist.
    • Could the NMOM (monomers in the unit cell) be calculated automatically? Is it an important parameter or is it only used to calculate the solvent proportion?
    • -> Conclusions regarding the chemical composition questions (see above):
      • Giving only the SOLVENT (+ solvent composition with SATM and PATM for other atoms that CHON) to RADDOSE would have been the simplest solution for the prototype. However, Gleb thought that it would be very inaccurate to go that way given that RADDOSE would assume that in this condition the molecule would be pure protein. Gleb recommended not to specify SOLVENT but rather the chemical composition of the non-bulk solvent part. This will be applied for the prototype.
      • Regarding the user inputs, Gleb suuggested to have user-friendly inputs rather than those expected directly by RADDOSE
        • NRES per monomer (or optionally monomer molecular weight) rather than NRES per chain (The postdoc of Elspeth Garman confirmed that NRES/chain = NRES/monomer, so there is no more ambiguity on this parameter.
        • (*NDNA NRNA and PATM are expected to be per monomer in RADDOSE, so it is OK)
        • NMON per asymmetric unit rather than NMON per unit cell
        • -> The recommendation will be to keep the parameters as close as RADDOSE expects (NRES, NDNA, NDNA, NMON, PATM for other atoms than CHON, SATM) and to translate the User-friendly inputs to the RADDOSE in the translater layer.

Output

  • Shall we limit the use case outputs only with those needed by BEST? -- MF
    • Having a full raddose plugin would be very interesting but the recommendation is to limit the output to what best needs for the prototype. We will implement a full raddose plugin in a second step.
  • The absorbed dose should be converted in Dose/second in the BEST plugin