Setting up an in silico experiment is generally as simple as following three easy steps:
1) Name Your Job
2) Select Your Protein(s) and Experiment type(s)
3) Select/Upload/Draw Ligand(s)
The majority of experimental options one would normally be required to select for the Support Vector Machine (SVM) or Molecular Docking software have already been pre-optimized for you based on our considerable experience and expertise with these programs. The SVMs we use come from peer-reviewed publications and were developed specifically for the protein they are listed. The objective is to make robust in silico experimentation available for experimental scientists who are not themselves computational chemists or bioinformaticians.
The protein/experiment selection sub-menu is organized by protein group/type. Under each protein group is a list of available experiment types. Expanding any experiment-type sub-menu will provide a list of the available specific protein or protein structures available to be used for that experiment. Selecting a higher-order checkbox will automatically select all lower level experiment(s)/protein(s). You may select any number of experiments and proteins to run at any given time, subject only to the limitations of your account type.
Once you have selected your experiment and protein of interest, you must select one or more ligands. You can chose from several publicly available ligands or load/draw your own. We provide a search function, allowing you to search for a particular ligand by name from our supply of public structures. You can upload your own ligand/ligands in SDF format, or you can use the interactive drawing function to draw a ligand of your choice. An SDF file can contain multiple ligands to be tested simultaneously.
Once you have selected your experiment, protein, and ligand simply click “Create New Job”, your experiment will be added to the computational queue, and you will be taken to the experiment dashboard. The progress indicator will inform you of the status of your job(s) and provides links to the results upon completion.
The progress in the area of Molecular biology has led to the generation of large amounts of data which is almost impossible to analyze and understand manually. Bioinformatics has aided in making many advances in the interpretation and usability of this data. Drug discovery and testing of experimental therapeutics is a rapidly growing avenue of structural informatics. With algorithms that predict structures, interactions and functions of biological molecules, it has become easier and far more practical to carry out initial testing and experimentation in-silico rather than going through the expensive and time consuming wet lab techniques. With our software “Althotas Virtual Laboratory”, we intend to facilitate researchers in the fields of drug design and discovery to easily simulate a docking experiment, predict potential substrates and apply chemical and physical calculations on their target drug-receptor complexes. In prediction of interactions between a drug and its receptor and computation of molecular features of a structure, machine learning and pattern recognition systems are used which apply Support Vector Machines. SVM calculations have been applied in prediction of protein secondary structure, cleavage sites, computation of molecular properties and determination of gene functions in the field of bioinformatics. In the Althotas Virtual Laboratory SVM prediction is being used to identify the drug targets while docking algorithms are applied to compute how a drug or ligand will behave when present in proximity of a certain receptor. The Althotas server has a friendly GUI and simple handling. It has joined SVM prediction techniques and docking experiments on one platform, where it can calculate functions and perform experiments on multiple ligands and receptors simultaneously and with high degree of accuracy.
Here we will demonstrate the functions and usage of Althotas server with an example experiment.
Once you have logged in to the Althotas server with your respective username and password your Dashboard will appear which will have the list of your previous jobs if any. For each of the enlisted jobs you can view the results, summary and/or download the results to your hard drive.
The Create new Job button will open a workspace where you will specify the job title, choose what type of calculations/experiments you want to carry out and specify the ligand for your experimentation. There are three categories of experimentations that can be performed; you can Calculate SVM predictions ,Dock Ligands and Experimental receptors and calculate different chemical and mathematical functions e.g. logP, atomic polarizability, no. of substituted benzene rings etc. In case of an SVM prediction the result will be presented as a numerical value and it will be stated whether the substrate is valid for the receptor or not, while a docking experiment will create a structural simulation of the two molecules interacting with eachother and compute the free energy required for the ligand and its receptor to interact in kcal/mols.
The Specify Ligand label allows you to either upload a file from your hard drive or user account, choose one from the list of ligands available on the Althotas server or Draw a ligand manually. Another option is to search multiple ligands in the Althotas server and then create a .sdf file with a unique name which will be saved to your account and can be accessed later from the ‘Upload ligand file’ option.
The Althotas virtual Laboratory can perform both SVM predictions and docking experiments with albumins, pgps and cytochrome p450s. The server is continuously expanding the list of the drug targets and adding more experiments to its virtual laboratory.
In this tutorial we will discuss an experiment of multiple SVM predictions and one docking example with an albumin molecule.
Click on the “Download ligand file”, type in or paste multiple ligands in the text box and specify a name for the file or you can search each ligand separately in ‘Search by name’ tab and choose from the ligand list. Once you have made a ligand list and named the file it will appear under the Select from available tab.For this virtual experiment we are computing SVM predictions for p-glycoproteins, Cytochrome p450s and albumins along with docking one albumin structure to each member of our multi-ligand list simultaneously.
For this example the ligand list contains 13 drugs that are, Aciclovir, Allopurinol , Amikacin sulphate, Amphetamine
Azapropazone, Bumetanide,Busulfan, Chlorodiazepoxide, Chlorothiazide ,Decitabine, Diazepam , Diiodisalicylic acid and Dopamine.
From the experiments list box, check SVM calculations for pgp, all SVM predictions available for cytochrome p450s that are CYP1A2, CYP2A6, CYP2D6, CYP3A4, CYP2C9 and albumins. For docking check one structure from albumin which is HSA_Fusidic_Acid_2VUF in this example. therefore this virtual experiment will tell us about the ligands that bind to albumin and the geometry of their binding, it will also explain which drug can be a pgp substrate and which cytochrome p450 plays a significant role in the drug’s metabolism.
After assigning a title for this job click on Create new Job button, the finished job will appear in your dashboard from where you can click on the “view results” which will open in the form of a table. Once SVM calculations and docking is done simultaneously, some columns will be illustrating the results of docking while others are specific to SVM predictions. Each row illustrates the output of docking and SVM against one receptor-ligand pair. Description of each column in the results table is given below:
Please note that the proceeding column numbers are valid only for the results of this example, for other jobs the number of result columns will depend upon the number of virtual experiments that you have computed on your data.
The first column of each row shows the structure of the ligand (i.e. Allopurinol Chlorothiazide ,Decitabine, Diazepam etc in this example)
The second column allows you to View or Save docking Geometry where you have the option of either saving the docking results in a .pdb file format or viewing the geometry of docking in a separate window. From the quick menu in the viewing window you can save the changes you make to the image, save the file as png image or open another structure from your hard drive. The select menu allows you to popout a control panel window where you can perform all the functions and edits on your structure.
The Measure menu can calculate angles, distances, torsions in the molecular structure.
In the control panel the “Select” radio button allows you to highlight the selected atoms, residues, chains, ligand etc. Similarly the append and exclude radio buttons will do their designated function. User can also change colours, width of lines and overall view of the structure by giving it shadow or fog etc. From the view controls you can dim the front or back portion of the structure to focus on one part.
The Generate tab in the control panel allows you to build ribbon structure of the molecules and then edit the width and thickness properties of helices, coils and arrows in the ribbon view.
The third column assigns a Mol id, Each ligand is given a specific molecular id which increases by one in case of experiments done with multiple ligands.
The fourth column gives the numerical value of SVM prediction i.e. 1, 0 etc while the next column explains whether the selected ligand (e.g.. aciclovir in the first row) can be a substrate for pgps, p450s or albumins or whatever the supposed receptor is. ‘0’ in the SVM result value column means that it can not be a substrate for the selected receptor while the value of 1 or greater means that it can be a substrate. This description is true for this particular example, for other virtual experiments the column of SVM result explains the significance of SVM’s numerical result.
The final column in this example before which estimations of free energy of docking starts states the similarity of the selected receptor to its ligand compound, which varies between a 0 to 1.
The last columns in the results table state the free energy of docking in kcal/mol for each of the potential bonding of the two molecules. The results are sorted from lower to higher energy values.
The results table gives you the option of sorting and filtering your data. For instance, By pressing the Simple grid button only the lowest free energy of docking is shown while all other values are filtered out. When the cursor is pointed on the heading of each column a drop down box appears from where you can sort the results in either ascending or descending order. User can also check or uncheck which result columns he wants displayed and which ones are to be omitted from the results.
In a nutshell, this virtual experiment has told us that the only possible substrate for pgp is the drug Amikacin sulphate. In cytochrome SVM experiments CYP1A2 is the only important factor in aciclovir’s metabolism. allopurinol is a substrate for all cyp450s except for CYP3A4, contrarily Amikacin sulphate only binds to CYP3A4 and no other CYP. Amphetamines and dopamine can not be a substrate of CYP3A4 and CYP2C9.
In case of Azapropazone, the cyp450s that play a significant role in the drug’s metabolism are CYP1A2,CYP2D6 and CYP3A4. The only possible receptor for Bumetanide is CYP3A4. while ,Busulfan, Chlorothiazide, Diiodisalicylic acid can be possible ligands for all cyp450s except CYP2D6 and CYP2C9 , Chlorodiazepoxide is a ligand for cyp450s except for CYP2A6 and CYP2C9, The targets for Decitabine can be CYP1A2 and CYP2A6 while Diazepam is a substrate for CYP1A2 and CYP3A4.
With albumin SVM predictions , the results show that binding site 1 can be pivotal in interaction between albumin and Bumetanide, Chlorothiazide and Diiodisalicylic acid, while Chlorodiazepoxide and Diazepam can be a substrate for binding site 2 of albumin.
In case of docking experiments with HSA Fusidic acid the lowest binding energy of -8.0 kcal/mol results with ligands like Diazepam and Azapropazone.
These deductions can help us in designing drugs with similar binding domains as indicated in the results of virtual experiments and provide useful information about the binding or docking reaction that is energetically favourable and therefore can be further tested by other lab techniques.