Manual |
The SOMO (SOlution MOdeller) module of UltraScan (US-SOMO) initially contained only a bead modelling utility that was originally developed by the Rocco and Byron labs, respectively at the Istituto Nazionale per la Ricerca sul Cancro (IST, Genova, Italy) and at the University of Glasgow (Glasgow, Scotland, UK). The original code was mainly written by B. Spotorno, G. Tassara, N. Rai and M. Nollmann. The SoMo bead modeling utilities in SOMO are based on a reduced representation of a biomacromolecule, starting from its atomic coordinates (PDB format, mmCIF compatibility is planned), as a set of beads of different radii, from which the hydrodynamic properties in the rigid-body frame can be calculated, after overlaps between beads are removed, using the Garcia de la Torre-Bloomfield "supermatrix inversion" (SMI) approach (García de la Torre and Bloomfield, Q. Rev. Biophys. 14:81-139, 1981). The reduced representation is afforded by grouping together atoms and substituting them with a bead of the same volume, appropriately positioned. Importantly, the volume of the water of hydration theoretically bound to each group of atoms can be then added to each bead. The overlaps between the beads are then removed in sequential steps, but preserving as much as possible the original surface envelope of the bead model. The method was fully validated and reported in the literature (Rai et al., Structure 13:723-734, 2005; Brookes et al., Eur. Biophys. J., 39:423-435, 2010; Brookes et al., Macromol. Biosci. 10:746-753, 2010). Among the main advantages of this method over shell-modelling and grid-based procedures are a better treatment of the hydration water and the preservation of a direct correspondence between beads and original residues. For instance, the latter feature could be used to include flexibility effects into the computations. Furthermore, by identifying and excluding from the hydrodynamic computations beads that are buried and thus not "in contact" with the solvent, a large span in the size of the structures that can be analysed with this method without loss of precision is obtained: originally, structures from 5K to 450K were successfully studied, and the steady increase in the available computer power have and is continuosly expanding this range .
Subsequently, we have also improved the original AtoB grid method (Byron, Biophys. J. 72:408-415, 1997), which was already included within US-SOMO, by adding the theoretical hydration, accessible surface area screening, and a better preservation of the original surface. The possibility of changing the grid size in the improved AtoB could be very useful to study very large structures and complexes.
Later on, in US-SOMO was added an alternative, at the time far more computationally intensive method of calculating the hydrodynamics based on the analogy that exists between certain hydrodynamic and electrostatic properties, ZENO (see Douglas, Some Applications of Fractional Calculus to Polymer Science, Adv. Chem. Phys. 102:121-191, 1997; Douglas et al., Hydrodynamic friction and the capacitance of arbitrarily shaped objects, Phys. Rev. E 49:5319-5331, 1994; Mansfield et al., Intrinsic Viscosity and the Electric Polarizability of Arbitrarily Shaped Objects, Phys. Rev. E, 64:61401-61416, 2001; https://zeno.nist.gov/). Then, in May 2014 an interface and an analysis modulus for the boundary elements method BEST [S.R. Aragon, A precise boundary element method for macromolecular transport properties. J. Comp.Chem., 25, 1191-1205 (2004); S.R. Aragon and D.K. Hahn, Precise boundary element computation of proteins transport properties: Diffusion tensors, specific volume and hydration, Biophysical Journal, 91:1591-1603 (2006)] were implemented within US-SOMO.
In 2015, a comprehensive study was conducted to compare various hydrodynamic modeling approaches (Rocco and Byron, Computing translational diffusion and sedimentation coefficients: an evaluation of experimental data and programs., Eur. Biophys. J. 44:417-431, 2015, Erratum http://dx.doi.org/10.1007/s00249-015-1058-1; see also Rocco and Byron, Hydrodynamic Modeling and Its Application in AUC, Methods Enzymol. 562:81-108, 2015). The methods tested were SoMo with computations using either the SMI or ZENO approaches, AtoB with 5 and 2 Å grid sizes and SMI computations, BEST, all under the US-SOMO implementation, and, externally, HYDROPRO (Ortega, A., D. Amorós, and J. García de la Torre. Prediction of hydrodynamic and other solution properties of rigid proteins from atomic- and residue-level models. Biophys. J. 101:892-898, 2011). The results indicated that, on average, BEST and HYDROPRO tend to underestimate the translational frictional properties by ~-3 and -4%, respectively, while SoMo using either the SMI or ZENO approaches overestimates them slightly less (~+2%). The best results using the SMI approach were obtained by AtoB with a 5 Å grid size, ~+0.5. However, a combination of SoMo bead models without overlap removal and ZENO computations performed even better, with ~0% average discrepancy and all results within ±4%, not far from the average experimental error of ±~3%. For these reasons, starting from the May 2015 US-SOMO release, this combination was directly offered among the bead modeling hydrodynamic computations options.
More recently, a new implementation of the ZENO method, completely rewritten in the C++ language instead of Fortran, with greatly improved serial performance and able to utilize the multi-core capabilities of modern processors, thus allowing computational times shorter by a factor of ~100, was produced at NIST (Juba et al., J. Res. Natl. Inst. Stand. Technol. 122:1-2, 2017). This new ZENO code was implemented into US-SOMO, but it could be distributed only with Linux system executables. This was due to the US-SOMO underlying architecture which relied on the Qt3 now obsolete libraries. A major rewriting effort was then done, to recode US-SOMO under the up-to-date Qt5 framework. This has allowed us to introduce a new bead model generating option, which we call "van der Waals (vdW) with overlaps". The vdW with overlaps method allows generating a bead model where each atom present in the PDB file (except, by default, H2O molecules) is represented by a bead whose radius is equal to the atom's van der Waals radius as listed in the somo.residue table (see below). If water molecules are associated to any particular atom in the somo.residue table, their volume is calculated and added to that derived from the atom's van der Waals radius, and a final bead radius is then recomputed. No overlap removal is performed, and the ZENO method could then be used to compute the hydrodynamic parameters. While this direct method as so far implemented matches slightly worse the experimental parameters of the test proteins (Brookes and Rocco, Recent advances in the UltraScan SOlution MOdeller (US-SOMO) hydrodynamic and small-angle scattering data analysis and simulation suite. Eur. Biophys. J. 47:855-864, 2018), it opens up some interesting possibilities, such as using structures explicitly hydrated by Molecular Dynamics simulations. The vdW with overlaps method is under active development.
Even more recently, a new computational method that could handle bead models with overlapping beads of different size has been presented (Zuk et al., GRPY: an accurate bead method for calculation of hydrodynamic properties of rigid biomacromolecules, Biophys. J. 115:782-800, 2018). GRPY (Generalized Rotne-Prager-Yamakawa) works in the same framework as the SMI method, while solving the long-standing issue of overlaps between beads of different sizes and providing robust estimates of the intrinsic viscosity and the rotational diffusion coefficients (the latter are not computed by ZENO), but is rather more computationally intensive. A non-parallel GRPY code is now made available within US-SOMO as an additional computational option, and a full implementation allowing multi-core parallel processing in underway (February 2021).
Importantly, the GRPY availability has allowed us to re-examine the performance of the SMI method for what concerns models with non-overlapping beads. While small differences were found for the translational diffusion properties computations, we are sorry to report that the SMI calculations of the rotational diffusion and the intrinsic viscosity did not match the state-of-the-art GRPY results for simple bead models (linear arrays, compact arrays, round arrays, all with equal radius) test structures, with up to 40% differences for the rotational diffusion properties. This failure directly stems from the inadequacy of the so-called "volume correction" that was originally developed by J. García de la Torre and collaborators and that is implemented in the SMI routines (see Spotorno et al., Eur. Biophys. J. 25:373-384, 1997; Erratum 26:417, 1997). While the García de la Torre group has subsequently introduced improvements for the computation of the rotational diffusion and the intrinsic viscosity (see García de la Torre et al., Improved calculation of rotational diffusion and intrinsic viscosity of bead models for macromolecules and nanoparticles. J. Phys. Chem. B 111, 955-961, 2007), it turns out that these can be applied only for non-overlapping beads of equal size. Therefore, from the February 2021 US-SOMO release, the SMI method hydrodynamic computations output will NOT report anymore the rotational diffusion properties and the intrinsic viscosity values. Users interested in these calculations can utilize the state-of-the-art GRPY method, that is, however, computationally demanding, or, limited to the intrinsic viscosity calculation, the ZENO method, that can handle large structures.
US-SOMO also includes a fully functional Small-Angle X-ray or Neutron Scattering (SAXS/SANS) simulator module, which works on either the original atomic structure, or on a bead model, and has enhanced experimental data processing capabilities. In the modeling area, several methods are offered for the computation of SAXS and SANS I(q) vs. q curves. Some of these methods require explicit hydration of the PDB structure(s), which should be presently externally provided. A pairwise-distance distribution function P(r) vs. r computation starting from a PDB structure is fully operational for both SAXS and SANS, and includes a graphical mapping utility to visualize which residues in the structure are contributing to specified distance ranges. An indirect Fourier transform Bayesian algorithm, based on the work by Hansen (Hansen, J. Appl. Crystallogr. 33:1415-1421, 2000; Hansen, J. Appl. Crystallogr. 41:436-445, 2008), has been implemented for the computation of the pairwise distance distribution function from SAS data.
In the experimental data processing area, a novel HPLC-SAXS data
processing utility has been implemented, which starts with the transformation of a time series of I(q) vs.
q frames into a series of time chromatograms I(t) vs. t for each q value. A check of the baselines, potentially revealing capillary fouling due to the accumulation of material on its walls, can then be performed, and corrections applied. In case of overlapping or not baseline-resolved peaks, Single Value Decomposition (SVD) can be applied on the original or baseline-corrected data, the latter after automatic back-generation of the I(q) vs. q frames, to identify how many components are present in the data. Global Gaussian analysis/decomposition can then be performed on the I(t) vs. t for each q value dataset, followed by back-generation of the I(q) vs. q frames for each Gaussian peak. Several improvements are present in this area from the June 2015 release, like an integral baseline evaluation/subtraction procedure, with immediate testing of the results in the I(q) vs. q space, the possibility of peak decomposition using non-symmetrical Gaussian functions, an improved treatment of concentration detector data, and a tool to evaluate the data-associated errors, when necessary, from the baseline fluctuations.
The Guinier analysis of experimental I(q) vs. q curves offers the determination of the overall z-average square radius of gyration <Rg2>z and of the w-average molecular weight <M>w from global Guinier, of the z-average square cross-section radius of gyration <Rc2>z and of the w/z-average mass per unit length <M/L>w for rod-like macromolecules, and of the z-average square transverse radius of gyration <Rt2>z and of the w/z-average mass per unit area <M/A>w for disk-like marcomolecules.
The batch operations module includes supercomputing access, with an interface to Discrete Molecular Dynamics (DMD) programs (Dokholyan, NV, Buldyrev, SV, Stanley, HE, and EI Shaknovich. Discrete molecular dynamics studies of the folding of a protein-like model. (1998) Folding & Design 3:577-587; Ding F, Dokholyan NV. Emergence of protein fold families through rational design. Public Library of Science Comput Biol. (2006) 2(7):e85). Starting from the May 2014 release, you will also find the implementation on a supercompute cluster of the boundary-elements hydrodynamic computations BEST [S.R. Aragon, A precise boundary element method for macromolecular transport properties. J. Comp.Chem., 25, 1191-1205 (2004); S.R. Aragon and D.K. Hahn, Precise boundary element computation of proteins transport properties: Diffusion tensors, specific volume and hydration, Biophysical Journal, 91:1591-1603 (2006)], and the relative interfaces in US-SOMO to set-up the analysis parameters and analyze the computations results.
Other features include a model classifier in which calculated parameters can be compared and ranked against experimental data, and a PDB editor.
The program main window contains an upper bar from which all the options governing its operations can be controlled, and a main panel for program execution. However, due to its high level of sophistication, properly setting all the available options can be non-trivial for the general user. Therefore, the US-SOMO module is distributed with pre-defined default options that should allow the direct conversion of a PDB-formatted biomacromolecular structure file into a bead model, and the computation of its hydrodynamic properties, without the need of accessing the advanced options menus. In particular, the SoMo approach is based on properly defining the atoms and residues found in PDB files, and the rules allowing their conversion into beads. The US-SOMO distribution includes the definition of all the standard amino acids, nucleotides, carbohydrates, and common prosthetic groups and co-factors, but this list is by no means exhaustive, and the need to code for "new" residues is not a remote possibility. As this operation can be demanding, notwithstanding the user-friendly GUIs governing it, the pre-defined set of options includes approximate methods to deal with either missing atoms within coded residues, and/or not yet coded residues. Starting from the May 2015 release, the default option is to generate a single bead for each non-coded residue using average parameters. When non-coded residues are found, a pop-up panel will alert the user and present as options (i) to continue with the approximate method; (ii) to skip non-coded residues (not recommended), or (iii) to halt operations and then take proper action like coding for the new residue. For coded residues with missing atoms, since most often this is due to lack of crystallographic data, the default option is now to use the complete residue's bead(s), appropriately positioned (again, a pop-up panel will warn of such instances and present the alternative skip (not recommended) or halt operations options). Obviously, there's no cure for completely missing residues, which will have to be built in the original structure for reliable results, since the structure should contain all residues and atoms that are present in the "real" macromolecule studied in solution. Therefore, for best performance all residues should be properly coded in the US-SOMO tables (see below).
SOMO Program: These functions control the execution of the US-SOMO program, whose progress is recorded in the right-side main window (in the pictures that follow below, excerpts of the messages during the model building and hydrodynamic computation phases starting from the 1HEL.pdb hen white lysozyme structure will be shown). They are divided in three subpanels controlling operations that deal with the primary PDB file (PDB Functions:), operations relating the generation of bead models (Bead Model Functions:), and the computation of the hydrodynamic parameters (Hydrodynamic Calculations:). Note that the buttons that can be actuated at a given stage of operations are identified with black labels, while not-actuable buttons are identified with red labels. This color scheme can be changed from the System Configuration menu accessisble form the Configuration pull-down menu in the top bar.
|
PDB Functions: log10 [A-]/[HA] = pH - pKa [A-]/[HA] = 10(pH-pKa) If one considers [A-] = I , the ionized fraction, then [HA] = 1 - I is the non-ionized fraction. Therefore: 10(pH-pKa) = I/(1-I)
Therefore, we can calculate the fraction of protons bound to each ionizable atom and the fraction of hydration waters associated with it as a function of the entered pH. From the proton counts, summation over the entire protein allows to calculate the overall anhydrous Molecular Weight and the net charge at the entered pH. The Henderson-Hasselbach equation is also used in an iterative way over all ionizable atoms to find the isoelectric point of the (bio)macromolecule under examination. These values are now reported in the progress window after loading a PDB structure. $ULTRASCAN/binfor 32 bit machines, and $ULTRASCAN/bin64for 64 bit platforms. You can get a copy of RasMol from http://www.bernstein-plus-sons.com/software/rasmol/ (recommended, there it's under active development), or from http://www.umass.edu/microbio/rasmol/, or from http://openrasmol.org/#Software.
|
As shown in the picture above, besides visualizing the structure, the HEADER and TITLE fields of the PDB file will be displayed in the progress window, followed by a list of identified disulfide bonds between CYS residues, if any. This is a feature that was already present but was never made operational until the February 2021 release. It's active now by default, but can be turned off from the PDB parsing pull-down menu, where the cut-off distance (default: 2.5 Å) can also be changed. Importantly, if unpaired CYS residues are identified, their name will be changed to "CYH", for which a special entry is defined in the somo.residue table, since some molecular properties slightly differ between free cysteines and disulfide-bonded cystines. The residues sequence in both three- and one-letter codes (both "CYS" and "CYH" are reported as "C" there) is then displayed, followed, as shown in the picture below, by a summary of each reside count, percentage, and its associated, pH-dependent theoretical waters of hydration and the corresponding global hydration (g/g, i.e. g water/g protein), and by a list of various molecular properties calculated for the model: number of disulfide bonds, number of free SH, molecular weight, molar volume, partial specific volume (vbar) calculated at the specified T and the one that will be used in the hydrodynamic calculations, anhydrous SAXS excluded volume, anhydrous and hdrated molecular volumes computed from vbar and the global hydration (using the water molecular volume as defined in the Miscellaneous Options panel), number of electrons, number of protons (pH-dependent), net charge at the specified pH, isoelectric point, and the average electron density:
|
If problems are encountered with the selected PDB file, like the presence of non-coded residues or missing atoms within coded residues, they will be reported in the progress window either as warnings or errors. Starting from the May 2015 release, if non-coded residues or coded residues with missing atoms are found,
a pop-up panel will appear warning of the occurence and offering the alternative options to continue using approximate methods, skip the whole residue, or stop the program execution, waiting for corrective action to be taken.
|
Bead Model Functions:
|
A fifth button, Grid Existing Bead Model, will operate the AtoB grid routine on a previously generated bead model. This button is not available until a PDB file has been processed with any of the bead modeling primary options (see above), or until a previously-generated bead model file has been loaded (see below). If this operation is launched, the "-a2bg" suffix is automatically added to the filename of the new bead model. Alternatively, you can load one or multiple previously-generated bead model by clicking on either the Batch Mode/Cluster Operation (see here) or the Load Bead Model File buttons from the menu. In these cases, and if the model(s) was (were) generated/saved in the US-SOMO format, the various settings/parameters used in model generation will be displayed in the right-side progress window. Note that you can decrease the number of beads used, and thus the resolution of the model, by applying a grid procedure on a previously-generated bead model with the Grid Existing Bead Model option (see above). This could be useful when large structures are analyzed, although using the improved AtoB routine on the original PDB file while increasing the grid size (Build AtoB (Grid) Bead Model) seems to produce much better results. By selecting different file types extensions, other type of bead models can also be loaded, like the old BEAMS-format models, or DAMMIN/DAMMIF-generated models. In this case, a pop-up panel appears requesting entering the partial specific volume and molecular weight of the model. The SAXS/SANS Functions button present in this subpanel will allow to perform SAXS-or SANS-related simulations directly on the currently loaded bead model. (see here).
|
Hydrodynamic Calculations:
|
A partial list of parameters can be seen in a pop-up window as soon as the calculations are completed by clicking on Show Hydrodynamic Calculations. The pop-up window will also list the solvent type, temperature and its associated density and viscosity, as set in the Hydrodynamic Calculations options module. A full list of all the parameters is also available as a text file, which can be opened from the results' pop-up window. Such a list from a previously analyzed model can be opened also from the Open Hydrodynamic Calculations File button. Warning: starting from the February 2021 release, SMI calculations will not report anymore values for the rotational diffusion parameters (like the Relaxation time tau(h)) and the intrinsic viscosity. Recent tests (2020; see the introductory history at the beginning of this Help section) have revealed that SMI calculations are not reliable for the rotational diffusion and intrinsic viscosity calculations. Users can get accurate values for these parameters by running GRPY or, for the intrinsic viscosity alone, ZENO.
The Select Parameters to be Saved button will open a pop-up window (see here) where characterizing/computed parameters can be selected for saving in a comma-separated file for easy import into spreadsheets. Selecting the Save parameters to file checkbox will generate such file, with extension .csv. The BEST button will open a pop-up window where the hydrodynamic computations results retrieved from a supercompute cluster run using BEST can be analyzed, as shown here. Finally, by pressing the Model classifier button, you will access a tool for selecting a best matching model among a series of models, by comparing their calculated hydrodynamic parameters with user-provided experimental values (see here).
|
The black-background bar at the bottom of the progress window will instead report the detailed advancement of some of the steps in the various phases, like the current slice and atoms (or beads) involved in the ASA routine, the iterations in the SMI method, the stage of the LAPACK routines for GRPY. For small structures using the SMI method, or for low number of MC iterations in the ZENO method, these numbers will be barely flashing by in the box, but for GRPY or for large structures using the other two methods they will allow a more in depth monitoring of the various stages. An estimated of the % progress in the hydrodynamic computations for all methods is instead presented and constantly updated as a blue segmented bar with a numerical value in the white-background space at the bottom of the commands side (in the image above, this indicator is shown at 100%). Operations can be halted at any moment by clicking on the Stop button. To avoid inadvertendly losing data, the Close button will not immediately close US-SOMO, but confirmation will be required in a pop-up window. |
Five pull-down menus are presently available to access the various US-SOMO options:
Lookup Tables
SOMO
MD
PDB
Configuration
From this pull-down menu, you can call four different sub-menus controlling
the four tables containing the definitions of the atoms and residues found in
PDB files, and their SAXS coefficients. More in detail, you can define/edit the hybridizations, atoms and residues that need to be interpreted as beads in the bead model generation. These parameters are collected in different tables that are used as the components from which the bead sizes and positions are calculated. PDB structures can then be converted to bead models based on the bead parameters defined here. For SAXS simulations you also need the atomic scattering factors coefficients (four or five exponentials plus a constant) and the associated excluded volumes.
|
SOMO:
From this pull-down menu, you can access various panels where you can set all the available options for different steps in the program. These options are saved in a system wide config file $ULTRASCAN/etc/somo.configEvery time you close the SOMO program, the currently defined options will be saved in $HOME/ultrascan/etc/somo.configwhere they will be reloaded from upon startup.
|
MD:
From this pull-down menu, it will be possible in the future to access two options panels controlling Brownian dynamics simulations:
|
PDB Options:
From this pull-down menu, you can access two panels controlling the options for parsing the PDB file and for the model(s) visualization by RasMol.
|
Configurations:
|
This document is part of the UltraScan Software Documentation
distribution.
Copyright © notice.
The latest version of this document can always be found at:
Last modified on March 18, 2021.