QUELO User Manual

Table of Contents

1. An Introduction to QSP Life
2. Logging in to QSP Life
3. QSP Life User Interface
4. Free Energy Perturbation (FEP) using QUELO 
5. Preparing a PDB file for use with QUELO
6.  QM Region Selection
7. QUELO Background / Whitepaper
8. Free Energy Perturbation (FEP) using QUELO: A Simple Tutorial

An Introduction to QSP Life

1.1 Understanding our software

The software you are about to use, QSP Life, represents a novel approach to carrying out calculations relevant to pharmaceutical discovery that integrates quantum mechanics (QM) to a degree never before possible. This software provides simple access to state-of-the-art QM tools that have been carefully optimized for use on expansive cloud resources and makes it possible to move beyond the limitations of traditional tools that use classical force fields. The QSP Life platform provides access to QSimulate’s QUELO, a state-of-the-art platform that makes possible–for the first time–Free Energy Perturbation (FEP) calculations using a QM representation of the ligand and surrounding
binding site. If desired, QUELO can also be used to run FEP calculations with only a traditional classical force field, taking advantage of cloud resources so that anyone can run these calculations without investing in local computer resources.

Calculations are set up via an easy-to-use portal. The calculations are performed on the backend using the resources of a Cloud provider, that grants high throughput capabilities to anyone, no matter what your local computing situation might be. QSP Life is implemented using the Software as a Service (SaaS) paradigm. In this paradigm, you interact with the program through a standard browser, such as Google Chrome, and all calculations are carried out externally. This paradigm has become increasingly popular over the past years for several reasons, including:

• There is no need to install software locally.
• The software is always up-to-date.
• You can take advantage of massive compute resources available through the Cloud without the need to build out specialized hardware locally.
• The tools can be run on any platform that supports a standard browser, including any laptop, a tablet, or even a phone.
• You do not need to be physically at work to run the software, you only need a web browser and an Internet connection.
• If you log off, or if your local computer crashes, is lost, or is taken down, your data and jobs are not affected.

• Open a standard browser.
• Navigate to the unique URL that has been created for your organization:
  •  https://(YOUR_ORGANIZATION).qsimulate.com
  • Replace (YOUR_ORGANIZATION) with the shorthand name QSimulate has provided to you.
• Enter your login credentials (provided once you purchase the software).
• Upload your data, choose what you would like to calculate, and run.
• Once finished, the results from a run can be downloaded via the platform.

Logging in to QSP Life

QSP Life is run through a standard Internet Browser. This platform has been developed and tested using the Google Chrome browser and that is the browser we recommend. However, you should be able to run from any modern browser (Chrome/Edge/Firefox/Safari/etc.). If you encounter any unexpected display issues using a non-Chrome browser, it is recommended you try using Chrome to check if this resolves the issue.

You may run your browser on any hardware that supports a browser. This includes laptops, tablets, and even phones (although the limited size of a phone screen will render the experience less than ideal). Because all calculations are run on the cloud, the CPU and memory requirements for your local hardware are nominal. If your hardware is typically capable of running a browser without issue, it will work with our software.

To access the platform, visit the following URL:

https://(YOUR_ORGANIZATION).qsimulate.com

where (YOUR_ORGANIZATION) is replaced with the shorthand name that QSimulate will have provided to you

2.1 Account creation & logging in

When you visit the landing page described above, you will be presented with a dialog for logging in:

If this is your first usage of the platform, click on Sign up to create your account. You will be prompted to enter your information (full name, e-mail, password) as well as a token:

The token is a unique key code that has been provided to the primary contact individual at your organization. Only the person(s) with the token can create new accounts. Either the contact person will share the token with you, or else that person will create an account for you.

Note: the token will only allow the use of e-mail addresses from the associated institution. For example, if the token was provided to the institution whose domain is “AAA.com”, then accounts can only be set up that correspond to e-mail addresses of the form name@AAA.com. This token is only required the first time a new account (for a new e-mail) is created.

Tokens are only provided after an agreement is in place between your institution and QSimulate.

Once your account has been created, you can log in using the credentials you specified for the account when you created it: e-mail and password.

2.2 Forgotten password

If you have forgotten your password, you can recover it by clicking on “I forgot my password,” which will ask for your name and the e-mail address you specified when you set up the account. Your recovery credentials will be sent to that e-mail.

2.3 Troubleshooting

If, after reading this section, you continue to encounter problems creating your account or logging in, you can contact us at the QSimulate support portal:

https://qsimulate-ticket.atlassian.net/servicedesk/customer/portals

QSP Life User Interface

3.1 The Task Table

Once you log in, QSP Life will present you with the following interface. This is the Task Table, which lists all of the calculations you have set up through the platform. Each time you want to set up a new calculation, you start by adding a new Task (for which you supply a name), and that Task will appear in this table.

If this is your first time using the platform, this list will be blank, as shown below:

Once you have created one or more Tasks (calculations), they will be saved and will populate the table when you return to the Task Table page, which you do at any time by clicking on the word “Home” in the upper left part of the page:

3.1.1 Task Table Contents

Within the table, for each Task you have set up, the following information is provided:

• Name: The name you assigned to the Task when you created it. You can rename a task, if desired, using the “Rename” button to the top right.
• Status: The status of the Task.
  • Staged: The FEP simulation options and input are still in the process of being defined, and the Task has not yet been submitted.
  • Running: The FEP Task is currently running
  • Stopped: The FEP Task was stopped by the user after it had been started, and before completion. Stopping and restarting of jobs is supported by buttons accessible through the page for the Task
  • Complete: The FEP Task has been run and completed successfully.
  • Failed: The FEP Task was run, but failed for some reason. Details on the reason for the failure can be found on the page for the Task.
• Updated: The time and date when the Task was last updated
• Created: The time and date when the Task was first created

All columns of the Task table can be sorted by clicking on the column header. Clicking on the Name and Status headers once will sort in ascending alphabetic order, and clicking on them again will reverse the sort. Clicking on the Updated and Created headers once will sort in the order Newest → Oldest. Clicking on them again will reverse the order of the sort.

3.1.2 Choosing a Task

Clicking on any of the Tasks in the list will take you to a page where you can either set up and run the job (if the Status
is Staged or Stopped) or else look at results and/or error issues (if the status is Running/Complete/Failed). The contents of the Task page itself will be described in the chapter on FEP.

3.1.3 Task Renaming

In front of each Task name is a check box. If you check the box in front of a single name, the Rename button at the top right of the table will become active, and allows you to rename that Task. Note that two tasks cannot have the exact same name.

3.1.4 Task Deleting

You can Delete one or more Tasks by selecting the check box(es) in front of the Tasks, and then selecting the Delete button at the upper right. The panel supports the multi-selection of boxes for the Delete operation. If you click one box, then Shift+Click a second box, all boxes between the first box and the second box will be simultaneously selected. When you click the Delete button, you will be presented with a confirmatory dialog to ensure you want to perform the Delete. Note that once Tasks are deleted, they cannot be recovered.

3.1.5 Task Cloning

You can “clone” one or more Tasks by selecting the check box(es) in front of the Tasks, and then selecting the Clone button at the upper right. The panel supports multi-selection of boxes for the Cloning Operation. When you clone a task, a copy of the task is created with the same name, pre-pended by the words “Copy of.” An example is shown below. A cloned Task copies over all the user input data and options. Once a cloned job is created, the user is free to modify the inputs of the job. For example, you could clone a calculation that was run using a “classical” force field, then change the option to use the “QM” force field. Or you could modify the input data set, or modify the mutation map.

Once you create a cloned Task, it is entirely independent from the Task it was cloned from. It can independently be run, renamed or deleted.

3.1.6 Creating a New Task

Below the Task Table, you will find the dialog that allows you to create a new Task. To add a new Task (calculation), type the name you want to give that task into the text entry region next to “Name” and then click the Add button. Task names must be unique. If you attempt to add a Task with the exact same name as a task that already exists, you will get the error “Task name already used.” When you add a new Task, the setup page for that Task will open immediately.

3.2 Navigation And Menu Buttons

Above the Task Table, you will find the Navigation display at the upper left (next to the QSP Life logo). And you will
find the Menu button at the upper-right.

3.2.1 Navigation

Navigation is displayed to the upper left of the panel, and starts with the name “Home.” If you are within a Task page, then the navigation will show as

Home > Task_Name

Clicking on the word Home from any page will take you back to the Task List view (the same view as when you logged in).

3.2.2 Documentation

Clicking on the Documentation button will bring you to an online version of the documentation for this platform.

3.2.3 Settings and Expert Mode

The settings button (shown above) opens a dialog where you can change various settings for your account: Display user name, password, and turn on Expert Mode:

• Email: In the settings menu, you can see the email address associated with your account, but this email address cannot be changed.
• New Name: You can change the display name associated with your account. This name does not affect your login.
• Password: You can change your password through this dialog. Enter your old password and new password. You must adhere to the password rules: 8-or-more characters, both lower and uppercase letters required, and your password must also include at least one number.
• Expert Mode: A key feature in this settings panel is the ability to enable “Expert Mode.” Expert Mode, which is turned off by default, will display a variety of additional options in the FEP setup panel. These options are not required for default behavior, but may be useful to advanced users. These options include the ability to modify the timestep, the amount of equilibration sampling, the extent of the periodic water box, the timestep, and the method used to derive classical force field charges. Note that if you turn on Expert Mode, it will revert to “off” once you log out of your current session.
  
  

3.2.4 Logout

If you click the logout button, you will be logged out of the platform and returned to the Login screen. (Note: A user will also be automatically logged out after a certain period of inactivity.)

Free Energy Perturbation (FEP) using QUELO

4.1 Overview

This panel provides the user with the ability to run relative free energy calculations, using the free energy perturbation (FEP) approach. Features of this panel include:

• Ability to run FEP using a classic molecular mechanics (MM) force field
• Ability to run FEP using a combined quantum mechanics (QM) /molecular mechanics (QM/MM) force field
  • Both the bound ligand and interacting protein residues in the binding region can be treated using QM
  • Up to several hundred atoms can be treated at the QM level without any sacrifices in sampling or the FEP
    approach
• Fully automated: A simple-to-use interface replaces complicated setup and workflow for the user

From the user standpoint, performing FEP using a QM/MM force field is nearly identical to performing FEP using a traditional classical MM force field. The only difference is that in the QM/MM case, the user needs to define the region of the system that will be treated using QM. This is simply done using options provided through the interface.

The FEP calculation is fully automated to make these calculations simple to carry out. Among the Tasks that are automatically handled on the backend when a calculation workflow is started are:

• Parameterization for MM regions
• Partitioning between the QM and MM regions (for QM/MM simulations)
• Creation of a periodic solvation lattice
• Superposition of ligands to a bound reference ligand
• Creation of an optimized matrix of calculations to run to calculate relative free energies for the full input set
• Analysis and compilation of results
• Parallelized distribution of Tasks

Below, the QUELO interface is described in detail.

4.2 The FEP Task List

When you enter the platform, you will be presented with the FEP Task List, a list of calculations (Tasks) that you have previously set up and/or run, as well as a dialog to create a new Task. Clicking on a Task will bring you to the setup/results page for that Task.

For more details on the FEP Task List, see the chapter “QSP Life User Interface.”

4.3 Expert Mode

A number of options (described below) are only shown in Expert Mode. The options shown in (default) Standard Mode are sufficient for most users to run a reliable simulation. If you need access to Expert Mode, that is accomplished via a toggle in the User Settings panel.

For more details on enabling Expert Mode, see the chapter “QSP Life User Interface.”

4.4 File input specification

When you click on a Task with Staged status in the FEP Task table, you enter the setup dialogs for that Task. At the top of the setup, you will need to specify the input files that are used to run the calculation. These files provide structural information on the protein receptor, as well as a set of ligands bound to the receptor for which free energy calculations will be carried out.

Note that after you submit a calculation to run, in some cases you may decide you wish to add additional ligands to the
compute set. This is possible, and the process for this is described in its own section (Adding Additional Ligands to a
Computed Set) later in this document.

An FEP calculation will calculate the relative free energies of binding of a set of ligands to a common protein receptor. Limitations of a standard FEP implementation require that both that the ligands be broadly similar to one another, and that we calculate relative binding energies among those ligands rather than absolute binding free energies.

To run a calculation, the following components must be defined by the user, using the Browse/Upload dialogs in this section of the panel.

4.4.1 Errors, Warnings, and Deleting an Input Structure

After an input file is uploaded, the contents of that file will be summarized below the Upload button. The summary will include the name, the SMILES string (except for the protein receptor), and an Information field, which will be populated with an informative message when there is a warning or error related to the upload. In the case of an error/warning, in addition to a message in the Information field, the NAME and SMILES strings will be colored red (rather than black) so that it is readily apparent when there is an issue with the input. There is also a red “X” at the right of each summary line, which allows you to delete the uploaded molecule, if desired.

4.4.2 Protein Receptor

This is the common protein receptor to which all the ligands that will be scored are binding. It is assumed that all ligands bind to the same receptor site, and that a single input conformation of the receptor binding site is suitable for all ligands.

The input format of the receptor is “PDB.” Care should be taken to ensure that the input PDB file is suitable for the calculation. See the chapter “Preparing a PDB file for use with QUELO” for detailed information on how this file must be processed before using it with QUELO. QUELO is capable of performing some processing Tasks to prepare a PDB file for use, but other Tasks may need to be carried out by the user before importing the file.

Issues that will need to be addressed before import include

• Removal of symmetry replicates from the unit cell
• Addition of missing loops near the binding region
• Ensuring that the residue naming convention is consistent with that expected by QUELO
• Possible removal of unnecessary cofactor molecules
• Possible identification of structurally important waters and removal of other waters

A properly prepared input protein receptor file is critical to obtaining high-quality results, and the user should read the PDB preparation chapter carefully.

4.4.3 Reference Ligand

The user must ensure that at least one ligand is pre-oriented properly in the binding site. For example, the user may obtain this ligand from a crystal structure of the complex or from modeling. This ligand must be supplied with 3D coordinates (SDF or MOL2), and it will be used as the reference binding conformation for any other ligands that need to be reoriented (see below). The reference ligand must include all atoms, including hydrogens.

Whether you use an SDF or MOL2 format reference ligand input structure, the structure must include a full description of topology information. This requirement ensures that the program properly resolves any structural ambiguities. If you wish to use a bound ligand in a protein/ligand complex structure as your reference ligand, you will need to create a file that contains only the ligand structure and upload it here, separately, to use it as the reference ligand, and you will need to convert the format of the ligand to SDF or MOL2.

The user must specify a name for the reference ligand that is being uploaded in the appropriate box (to the left of the
Upload button). If this name is not supplied, the Upload button will not be activated.

4.4.4 All Ligands (excluding reference)

The user can specify any number of additional ligands for which FEP calculations will be performed. These ligands are provided as 3D structures in either SDF or MOL2 format.

You can supply more than one input file of input molecules. After one file is imported, clicking Browse and then Upload for additional files will append the molecules in the subsequent files to the bottom of the list.

For each molecule in the input, a Maximum Common Substructure (MCS) alignment with the reference ligand will be carried out to align the 3D structure with the reference ligand.

After the input file for this section is uploaded and parsed, the names and canonicalized SMILES representations for all the ligands will be shown in a table below the input dialog line. For each input structure, the name is taken from the input file (if provided) or is otherwise the LNNN name auto-generated by the platform. If you click on the red X at the end of any input ligand line, that ligand will be removed from the input (a confirmatory pop-up will appear when you click on the X to ensure you don’t inadvertently delete a structure). Additionally, you will find a red “X” in bold font at the top of the column of red X buttons. If you wish to delete all the added ligands in this section at once, click
on this bolder X.

Note that no two ligands can be identical. If identical ligands are found in the input, these will be annotated as errors in the Information field and shown in red.

You must remove (using the red X) all problematic ligands before you can proceed with the calculation.

Expert Mode Options for this section are highlighted in the sample figure by red boxes. These options are only presented when you have enabled Expert Mode. They are described below:

4.4.5 Generate all stereoisomers for additional ligands with undetermined chiral centers (Expert Mode)

• Ticbox checked: For input structures (other than the reference), if there are atoms with ambiguous stereochemistry, all possible stereoisomers will be generated.
• Ticbox not checked: If an input structure with ambiguous stereochemistry is encountered, it will result in an error, which will be reported to the user in the Information field for that ligand.

4.4.6 Use 3D Alignment (Expert Mode)

Default: ON Every additional ligand input needs to be aligned on the reference ligand before an FEP calculation can proceed. (An exception is when the user specifies that all ligands have been pre-aligned before input, see the options section). The default approach to alignment uses a topological approach to maximum common subsstructure (MCS). Better alignement can often be obtain using MCS metrics based on 3D structural information. If you enable this option, 3D structural information is used in the MCS alignment.

4.4.7 Use Star Map (Expert Mode)

Default: OFF By default, a mutation map that identifies closed mutation cycles will be generated. Such a map can be beneficial by both providing better estimates of calculation errors and by reducing certain errors. However, there is a significant computational cost to such a mutaiton map (with typical maps increasing the number of mutations carried out by a factor of two or more). Sometimes, for efficiency, speed, or cost, a user may wish to instead carry out a “Star Map” calculation, where all mutations are connected to common residue by a single mutation (edge). The total number of mutations carried in this case is N-1, where N is the number of ligands. The center of the star map will be the reference ligand, as specifed during input.

4.4.8 Max #Charge-Changing Mutations (Expert Mode)

Defines the maximum number of connections between ligand nodes in the mutation map that have differening formal charges. The default is a maximum of two connections. This value is described in more detail in the description of the mutation map below.

4.4.9 2D Ligand Viewer

A 2D representation of any ligand that has been imported can be viewed by clicking on the row for that ligand.

4.4.10 Rotate Bond (Expert Mode)

In the 2D ligand viewer panel (above), if the user is in Expert mode, it is possible to define a bond for rotation. The rotatable bond, if defined, is a bond for which both the input geometry and a second geometry, rotated 180 degrees about that bond from the input geometry, will be used during the free energy workflow. (If the Rotate Bond values are left blank, this rotation will not be performed). Note that rotations about bonds in the molecule are naturally possible during the molecular dynamics simulation. But for some molecules, there may be a bond for which the rotational barrier is high and so the molecule will not be able to sample sufficiently between conformers separated by that high barrier. In situations where it is ambiguous which conformer is more likely, the Rotate Bond function can be useful.

Start and end are set to the atom numbers (see in 2D plot) defining the rotatable bond. Rotation will be performed with the atoms on the “End” side of the bond moving, relative to the fixed positions on the “Start” side. A second conformed is generated by a rotation of 180 degrees about the rotatable bond. This results in two geometries: The starting (input) geometry, and a second geometry reflecting a 180-degree rotation about the selected bond.

The Mutation Map (below) will be generated as normal. But when a rotatable bond has been defined for a molecule, all mutations (edges) to/from the node corresponding to that molecule will be carried out twice, once to the starting conformer, and a second time to the 180-degree rotated conformer. Subsequently, the two free energies of mutation for each edge connecting to the node with two conformers will be Boltzmann averaged.

For a molecule where an additional conformer has been defined using the Rotate Bond option, two molecules will subsequently be associated with the molecule (node). The first retains the name as read from the input ligand file (or auto-assigned by the program if no name was given), INPUT_NAME. The second will be named INPUT_NAME-alt. For example, in the case where rotable bonds are defined for two ligands with the names “ref” and “6e”, the mutations table (described in a section below) would list the mutations list as:

4.5 Mutation Map

Once the file uploads are complete, you can create the mutation map. The mutation map creates a set of connections (known as mutations or “edges”) among the ligands that define the set of FEP calculations that will be used to evaluate the relative free energies of binding for the set of ligands. The Mutation Map must be determined before continuing with the calculation. Once all the input files have been supplied, press the Generate Mutation Map button that appears on the bottom right of the File Uploads section. An automated iterative optimization is carried out to define the mutation map. If the user wants to make changes to the map generated, that is possible using the editing tools (see Mutation Map Editing Below).

The mutation map has three goals

• Evaluate similarity among the ligands, and create mutation paths between ligands that are closely related (for which relative free energy calculations tend to converge more readily)
• Ensure that there is at least one pathway that connects every input ligand to the other input ligands
• Add redundant mutation paths that can help reduce the overall error in the calculated free energies

A sample map, for eighteen input ligands, is shown above. The map consists of nodes (shown as circles) and edges (lines that connect the nodes). Each node corresponds to one of the input ligands as defined by the user. The character string next to each node is the name of the ligand (see the Name field of the ligands table). Each edge is a mutation that will be carried out by the program. The width of the edge reflects the similarity, with thicker edges being more similar, and thin edges being less similar. If you move the cursor over any edge, a number will appear that provides a similarity score, which can range from 0.0 (completely dissimilar) to 1.0 (identical).

You can enlarge/reduce the size of the map using the scroll wheel of your mouse. Enlarging the local map is sometimes useful to better see the varying widths of the edges, or to better read the names next to the nodes.

Very low similarity indices (especially 0.0-0.4) should be examined carefully to ensure that the ligands connected by the edge are acceptably similar to make it possible to determine a reliable free energy. Note that there will usually be more edges (free energy mutations to calculate) than input ligands. As noted, the additional edges are to reduce the overall error in the net calculated relative free energies.

Hovering the cursor over any node will present the 2D representation of the ligand corresponding to that node.

The user should evaluate the mutation map carefully before proceeding, ensuring both that the edges make sense, and that the 2D representations are as you expect for the input ligands. Once you are satisfied with both, you can move on to specifying Simulation Options and then running the simulation.

4.5.1 Mutation Map Progress Bar

While the calculations are being performed to produce the mutation map are being carried out, a progress bar will appear at the top of the Mutation Map section of the panel. When the mutation map determination is finished, the progress bar will disappear and the map will be shown.

4.5.2 Mutation Maps with Formal Charge Changes

In the case where all the ligands have the same formal charge (e.g. +1, 0, -1, etc.), then the mutation map is generated as described above. If the ligand set includes ligands with a mixture of formal charges, then in addition to topological and conformational similarity, we must also consider the charge change in optimizing the mutation map. In this case, the process followed is:

• Separate ligands into sets based on their formal net charge
• Generate individual mutation maps for each group of formal net charges
• Also generate a mutation matrix for all ligands, ignoring formal charge (full set)
• Edges from the full set that connect the charge-based sets are introduced (on the basis of similarity and molecular weight) to ensure that there is edge connectivity for all nodes in the map (regardless of charge)
• By default, the maximum number of edges connecting nodes with differing charges is 2. This default can be changed in Expert Mode. In Expert Mode, the field where you can change that value appears to the left of the Generate Mutations button
  

Changing the Maximum Number of Charge Changing Mutations will have no effect if all ligands have the same formal
net charge.

4.5.3 Mutation Map Editing

Using the editing tools, it is possible for the user to clean up the map presentation on the screen and add or remove edges (mutations) from the map, as described below.

• Prettify: This button will automatically adjust the positions of the nodes on the screen to improve the visual presentation. Use this if nodes are overlapped, or if you have adjusted the map edges via editing and want a better view. The Prettify button does not change the underlying simulation matrix. Clicking this button won’t change what mutations will be carried out–it merely provides a nicer presentation of the matrix on the screen.
• Download: This button allows you to download the current view of the mutation map in the .SVG format.
• Edit: This button puts the mutation map in Edit mode. In Edit mode, you can modify the edges (lines) that connect the nodes (ligands) together. These define the set of mutations that will be carried out. If you click on the Edit Edge Button, two additional buttons, “+” and “-” appear at the lower right region of the region. Click on the “Done” button at any time to save the edits you have made and go back into View mode. You can click on Edit again, after “Done”, to make more edits, if desired.
• Stop: The Edit button changes to the Stop button when you enter Edit mode. Clicking on the Stop button commits any changes you have made during editing, and returns to the View mode.

To remove an edge, click on the center of that edge. The color of the edge will turn red. Then click on the “-” button to remove that edge. If you decide you do not want to delete that edge after clicking on it, simply click on another element in the map (another edge or a node), and the previously-selected edge will be de-selected. (You can also click on the Stop Editing button to achieve the same thing).

To add an edge, you need to click on two nodes you want to connect, then click on the “+” button to add the edge. The order in which you selected the nodes is important: The mutation direction will be from the first node you select to the second node you select. When you select a node, it will turn red. (If you want to de-select that node, just click it a second time). You cannot add an edge to two nodes that are already connected by an edge. When you click on the first node in Edit mode (Node_1), that becomes the root node for all additions in this edit session. Clicking on a second node (Node_2) then on the “+” button adds an edge between those two nodes (Node_1-Node_2). Then, clicking on a third node (Node_3) and the “+” button will add another edge between the first node and the third (Node_1-Node_3). If you wish to pick a different root node for editing, click the “Done” button, then click “Edit” again to start a new set of edits. (The previous edits are saved and retained when you click Done).

When editing edges, you need to ensure that every node is somehow connected to every other node through a pathway of edges. More generally, to reduce the net statistical errors in the calculated binding energies, every node should be part of at least one closed loop.

If you create singletons (nodes with no connections) or islands (a group of nodes connected to each other but not connected by any edge to other another island of nodes), the automated workflow will fail. If the ligand set is such that there is no obvious way to connect certain nodes (because islands of nodes are too dissimilar to each other), you may need to set up multiple simulations, one for each island of nodes.

When you are satisfied, click the “Stop Editing” button. If you make a mistake and wish to return to the default automatically generated map, you can re-generate the map by clicking the “Generate Mutation Map” button.

If you make a mistake during editing and wish to revert to the starting map, click on the Generate Mutation Map button at the bottom of the file input section. This will delete the current mutation map and replace it with one auto-generated from the input files.

4.5.4 Star Mutation Maps (Expert Mode)

If the user has chosen the “Generate Star Map” option in the previous section, then instead of the closed cycle map described above, a star map is generated, where one edge connects the reference ligand to every other ligand in the set, as shown below:

This simple map requires the minimum number of mutation edges to be calculated and can be useful to save time or compute resources, although at the possible expense of increased overall error. The reference ligand, as specified during input, appears at the center of the star map. All other functions (Prettify/Download/Edit) operate as described for the standard Mutation Map, above.

4.6 Adding Additional Ligands to a Computed Set

On occasion, after starting or completing a simulation, you may decide you wish to add more ligands to the set. While you could run an entirely new simulation specifying the larger set, if you have already run calculations for some members of the expanded set, you may wish, instead, to simply add the extra mutations to those already evaluated to save on computational cost.

You can add additional ligands to a calculation that has either been paused (stopped) or else a calculation that has finished. You cannot add ligands to a job that is in “queued” or “running” status. If you have submitted a job, and the job has not yet been completed, and wish to add additional ligands, click on the Stop button, wait for the status to change to Stopped, add the ligands, and then Restart the job.

To add ligands to a stopped or completed job, you simply upload additional ligands via the All ligands (excluding reference) dialog, as described above under File Input Specification. The additional ligands will appear in the list of uploaded structures that appear under the upload dialog, at the end of the list.

After uploading additional ligands, if you subsequently wish to delete any of the added ligands, you can delete them from the list using the red “X” buttons on each line of the uploaded ligands set. (Be careful not to delete ligands from the original set; if you delete one of those ligands, any calculations related to that ligand will be deleted from the project and cannot be recovered without uploading that ligand again and re-running).

Once you have uploaded the additional ligands, you will see a pop-up that warns you that you will need to regenerate the mutations map, using the Generate Mutation Map button. Clicking on this button will generate a new map that includes the additional ligands, but retains the map that was generated prior to adding the additional ligands. Edges and nodes that include the newly added ligands will be shown in blue. Below, as an example, is the original map generated for a set of ligands, and the updated map that was generated after adding additional ligands. Note that the map for the original ligands may appear slightly different in the new map, but the connections (edges) will be the same. In this example, note that the edge between ligands 17 and 1q is drawn pointing downward in the original map, and it is drawn pointing upward in the map with the additional ligands. This has no effect on the calculation but aids in presenting a clear view of the map with the added vertices/edges.

By forcing the original map to be retained, the number of additional edges that is generated through this process is minimized and computational cost is kept in check. The resulting map can be edited just as described for the original map, and you can both add and remove edges. Note that if you attempt to remove an edge for a calculation that was already carried out in the first part of the calculation (before adding ligands), the mutation for that edge will be permanently removed and cannot be retrieved. A warning will appear to avoid inadvertently deleting edges for the previous calculations.

Note, also, that because the original map (from before adding additional ligands) is retained in any expanded map that results from adding additional ligands, it is likely that the overall map that is generated when adding ligands to a pre-existing calculation will be different from the map that would be generated if the full set of ligands (from the two steps) had been specified from the beginning. The original map generation attempts to optimize similarity connections while limiting a total number of edges. The resulting edges can and will differ depending on the set of ligands provided. Typically, the edge structure of a calculation performed without adding additional ligands later will be overall better
optimized. That said, the map generated in two steps will still connect all vertices with at least one edge.

Some other requirements when adding additional ligands:

• Do not change the map type (standard vs star map). Whatever option you chose originally should be maintained.
• Do not change the Simulation Options (below).

After you have uploaded the additional ligands and generated the map (and edited the resulting map, if desired), you can click on the Resume button (below) to restart the simulation for the revised map/ligand set.

The upside of simply adding to an existing calculation is that you reduce the number of new edges that need to be calculated. The downside is that in order to reduce the number of edges added, the resulting map may not be as well optimized as if you had run everything at once.

4.7 Simulation Options

The options displayed will depend on whether you are running in Standard Mode (default) or in Expert Mode. As noted earlier, Expert Mode, if desired, is turned on in the User Options panel. The standard options will be described first. Expert Mode options will be described in the subsequent section. You will only see one of the FEP Options sets, depending on the status of the Expert selection toggle.

4.7.1 The Overall FEP Workflow

The overall FEP workflow, which places many of the options described in the following sections into context, is shown in the following diagram. Here, you can see how the pre-FEP MD equilibrated structure feeds into each lambda simulation, where it is used as the starting point for additional equilibration and then data collection.

4.8 FEP Options (Standard)

A limited number of options are provided here. Reasonable defaults are provided for most options related to FEP and are not shown here. The remaining options include:

• FEP type: Defines whether this FEP simulation will use only a classical MM force field, or whether this simulation will incorporate a quantum region around the ligand, and use a QM/MM force field.
  • Classical: Use a classical force field for the entire system
  • Quantum: Use a QM representation for the ligand and surrounding residues and a QM/MM force field.
• #windows: Number of windows that connect the two ligands for which a free energy difference is calculated.
  Convergence is improved by breaking up the free energy simulation into a set of more similar lambda intermediates and then summing the free energies for those intermediate states to get the total free energy. The number of lambda points connecting the endpoints is #windows+1. The default is #windows = 12 for Classical, and #windows = 13 for Quantum. For Quantum, the extra window is inserted between the two lambda points near the endpoint to improve convergence.
• Simulation time (ns): The amount of molecular dynamics (MD) sampling done for each FEP window to obtain a converged value of the quantity that determines the free energy change for the corresponding lambda window. The default value is 5.0ns for a Classical simulation, and 2.0ns for a Quantum (QM/MM) simulation.
• Reset to Default: Reset all options to their defaults.
• Start Simulation: This option will start the FEP simulation. This option only appears for FEP_Type=Classical.
  For FEP_Type=quantum, additional options related to the quantum region need to be set after system preparation. If you need to set the calculation to pause after the Preparation stage for a classical simulation, tick the box next to Stop after preparation (found above this button).
• Start Preparation: This option will start FEP preparation. This option only appears for FEP_Type=Quantum.
  For FEP_Type=classical, stopping after preparation is not typically necessary and the preparation and simulation steps are combined.
• Stop after preparation: Pauses the workflow after the preparation step. This is the default (and can’t be changed) for FEP_Type=quantum, because the user must subsequently define the quantum region. For FEP_type=classical, the user can force a pause in the workflow by clicking this box.

4.9 FEP Options (Expert)

If you have enabled the Expert Mode in the User Settings panel, then a larger number of FEP options are accessible.
These options will appear below the Mutation Map.

• FEP type: Defines whether this FEP simulation will use only a classical MM force field, or whether this simulation will incorporate a quantum region around the ligand, and use a QM/MM force field.
  • Classical: Use a classical force field for the entire system
  • Quantum: Use a QM representation for the ligand and surrounding residues, and a QM/MM force field.
• Pre-aligned Ligands: By default, additional ligand that is input (except the reference ligand) will be automatically reoriented to best overlay the reference ligand, using MCS. If you tick this box, then no reorientation will be performed for additional ligands. This option requires that the user ensure they have been pre-oriented to the reference ligand. If the user chooses this option and fails to pre-orient their ligands before importing them, the resulting calculation can fail to converge or fail entirely. With this option toggled on, the single and dual topology regions of a pair of molecules are determined by comparing atomic coordinates. Any atoms closer than 0.1 Angstrom will be considered part of the single topology region.
• Timestep Length (ps): The integration timestep in picoseconds. Default is 0.002ps. Generally, you should not exceed this value, but lower values are acceptable (though lower values will require more computational timesteps to achieve the same effective length of simulation).
• Random Number Seed: A random number that will be used to create a chain of random numbers. These random numbers are used to set the initial velocities for the system. Using a different random number will create a different set of velocities and a different overall trajectory. A value of “-1” uses the default random number seed, which is based on the current Unix time when the job is submitted. Submitting two jobs with the default value of -1 and no other differences would generate two trajectories that vary due to their stochastically-generated initial velocities. If you wish to be able to exactly reproduce the same results when running the same job twice, you need to specify a (non-default) random number seed. E.g. running the calculation twice with a random number seed of 32177 would give identical results.
• Pre-FEP Equilibration (ns): The total amount of MD equilibration to perform before starting the FEP runs, in nanoseconds. This value does not include any MD requested by Heating Duration and Pressurization Duration (below). The default is 0.25ns (250ps). This value also does not include any additional QM/MM equilibration (below) that is performed in the case of a Quantum simulation.
• Heating Duration (ns): The temperature will be ramped up from the initial temperature of 0K to a target temperature of 300K over the specified number of nanoseconds of the simulation. This Heating Duration ramping time is not included in the total Pre-FEP Equilibration time (above). The default is 0.1ns (100ps). If no temperature ramping is desired, then set the Heating Duration to 0.00.
• Pressurization Duration (ns): The pressure will be ramped up from the value calculated from the input system to the Target Pressure of 1atm over the specified number of nanoseconds of simulation. This Pressurization Duration ramping time is not included in the total Pre-FEP Equilibration time (above). The default is 0.1ns (100ps). If no pressure ramping is desired, then set this value to 0.00.
• QM/MM equilibration (ns): For a QM/MM simulation, additional pre-FEP equilibration is generally applied to more thoroughly allow the system to reflect the force field. The QM/MM equilibration is performed using the QM/MM force field that will also be used for the FEP calculation. QM/MM equilibration is not applied for “Classical” mode. For QM/MM, total equilibration is therefore: (Pre-FEP Equilibration) + (QM/MM equilibration) The default is 1.0ns.

The temperature heating ramp is performed before the pressurization ramp, after which the Pre-FEP Equilibration MD is performed. Total FEP Pre-Equilibration simulation time will be:

(Heating Duration) + (Pressurization Duration) + (Pre-FEP Equilibration) + (QM/MM Equilibration).

(Default is 450ps total for a classical simulation, 1,450ps for a QM/MM simulation).

The remaining options define the number of intermediate “lambda” windows, and the amounts of equilibration and data collection performed at each window. The defaults are set to widely-accepted values used in literature studies. A full free energy calculation consists of two separate calculations, one for the ligand bound to the receptor, and a second for the ligand free in solution. The values defined here will be used for both separate calculations.

The FEP simulation at every lambda point starts with the equilibrated system as generated according to the Pre-Equilibration Options above. Additional equilibration is performed at that lambda point, using a potential energy function that reflects that specific lambda point. The amount of additional equilibration to perform is defined by the “Equilibration (ns)” parameter below.

• #windows: Number of windows that connect the two ligands for which a free energy difference is being calculated. Convergence is improved by breaking up the free energy simulation into a set of more-similar lambda intermediates and then summing the free energies for those intermediate sets to get the total free energy. The number of lambda points connecting the endpoints is #windows+1. The default is #windows = 12 for Classical, and #windows = 13 for Quantum. For quantum, the extra window is inserted between the two lambda points near the endpoint to improve convergence.
• Random Number Seed: A random number that will be used to create a chain of random numbers. These random numbers are used to set the initial velocities for the MD used in generating the FEP statistics. Using a different random number will create a different set of velocities and a different overall trajectory. A value of “-1” uses the default random number seed.
• Equilibration (ns): The number of nanoseconds of MD equilibration to perform at each lambda point before data collection starts. The default is 0.5 ns.
• Simulation time (ns): The amount of molecular dynamics (MD) sampling to perform for each FEP window to obtain a converged potential energy trajectory used to determine the corresponding free energy change for that window. The default value is 5.0ns for a Classical simulation, and 2.0ns for a Quantum (QM/MM) simulation.
• Reset to Default: Reset all options to their defaults.
• Start Simulation: This option will start the FEP simulation. This option only appears for FEP_Type=Classical. For FEP_Type=quantum, additional options related to the quantum region need to be set after system preparation. If you need to set the calculation to pause after the Preparation stage for a classical simulation, tick the box next to Stop after preparation (found above this button).
• Start Preparation: This option will start FEP preparation. This option only appears for FEP_Type=Quantum. For FEP_Type=classical, stopping after preparation is not typically necessary, and the preparation and simulation steps are combined.
• Stop after preparation: Pauses the workflow after the preparation step. This is the default (and can’t be changed) for FEP_Type=quantum, because the user must subsequently define the quantum region. For FEP_type=classical, the user can force a pause in the workflow by clicking this box.
  
  

4.10 FEP Preparation

4.10.1 Ligand Alignment

The first stage in preparation is to accurately align the ligands to allow the automated determination of topological correspondences required for FEP calculations. At this stage, parameter assignments for the MM region and other preparatory work is performed. The progress of this preparation portion of the workflow is indicated by a progress bar, as shown.

4.10.2 QM Region Specification

The QM region specification options are described in more detail in the subsequent chapter, “QM Region Selection.” Here, a simple description of the options is provided.

If you have set FEP Type = Quantum in the FEP options, then an additional panel will appear below the FEP Options section. If you are running an FEP Type = Classical simulation, the entire system is treated using MM, and this section is hidden.

In this section, the user can specify the extent of the region to be treated at the quantum mechanics level. The MD calculations are carried out using a QM/MM implementation, where the core residues (the ligand and chosen surrounding residues) are treated with the higher accuracy QM approach, and the remainder of the system is treated using a classical molecular mechanics force field. Note that explicit water solvents and ions are always treated using MM.

For QM/MM simulations, the ligand must always be included in the QM region. The ligand is represented by the keyword “ligand” (lowercase, no quotes) in the definition box.

Residues included in the QM region are filtered by the maximum number of atoms in that region. The maximum atoms filter is applied when generating an automated definition of the QM region (Suggest a QM region), along with a cutoff distance of 5 Angstroms. The distance between the ligand and a protein residue is defined as the shortest distance of any atom of the ligand to any atom of that residue. If that distance is less than 5 Angstroms, the residue is a candidate for being included in the suggested QM region.

• Region Max Atoms: The maximum number of atoms that will be included in the QM region if the region is automatically generated using the Suggest a QM Region button (below). The atom cutoff is implemented on a residue basis, so that any residue is either entirely within the QM region, or it is excluded from that region. The default is 100 atoms. Increasing the number of atoms in the QM region will increase the computation required for the simulation to finish.
• Selection Algorithm: [Simple/Optimal] This selection applies when using the automated “Suggest a QM Region button (below). In both cases, an initial set of protein residue candidates for inclusion in the QM region is generated based on a cutoff distance of 5A from the ligand. A ranking value is assigned to every residue candidate, and then residues are removed from the candidate pool, based on rank, until the Region Maximum Atoms limit is met.
  • Simple: The ranking value for each protein residue in the candidate set is based on the minimum distance of any atom of the protein residue to any atom of the ligand. A higher ranking value corresponds to a larger distance. Residues are removed from the candidate pool in order of ranking (high to low).
  • Optimal: The ranking value for each protein residue in the candidate set is based on both the minimum distance of any atom of the protein residue to any atom of the ligand and whether the protein residue makes specific close-distance polar interactions with the ligand. A higher ranking value corresponds to a residue making less significant interactions. Residues are removed from the candidate pool in order of ranking (high to low).
• Region Max Atoms (Charge)
  The atom maximum here applies only if there are mutations between residues that have differing net formal charges. If there are no mutations between nodes with differing formal charges, then this option will be greyed out. The QM region for mutations where a formal charge change occurs is selected via a different approach than the remainder of the mutations. For formal charge change mutations, the chosen QM region is informed by QM-based polarizability calculations on the binding site. Typically, a somewhat larger maximum number of atoms is required for these mutations. With this option, you can set the maximum number of atoms for charge change mutations (or accept the default).
• Algorithm (Charge)
  The algorithm selector here applies only if there are mutations between residues that have differing net formal charges. If there are no mutations between nodes with differing formal charges, then this option will be greyed out. As with the standard Selection Algorithm, there are two options, Optimal (default) and Simple. The operation of these options is described above, under “Selection Algorithm”.
• The text input area: In the middle of this region, you will find a text input box. You can specify residues that are to be included in the QM region in this text box using the format described in the chapter “QM Region Specification.” This box can be populated either manually, by writing in the definitions on your own, or else automatically by pressing the “Suggest a QM Region” button. The ligand must always be included in the QM region, using the definition “ligand” (lowercase, no quotes). If you use the Select a QM Region button, the ligand will appear at the top of the suggestion text automatically.
• Suggest a QM region: Auto-populates the text input area with a set of residues that meet the criteria described Selection Algorithm (above). The user can modify the list once it is generated (adding or subtracting residues).
• Validate: This button allows the user to validate the set of residues that are specified in the text input area above the button. Validate will ensure the residue definitions correspond to residues in the input PDB file. Validate will also add capping residues, if necessary, to ensure proper QM/MM boundaries for the calculation. If you use the Suggest a QM Region button and do not modify the contents of the definition box, the contents are automatically validated, and the Validate button is inactive. Only if you change the contents of the definition box will the Validate button be active.

Visualize: Creates a pop-up window that shows the selected QM-region residues in the context of the protein/ligand complex. The reference ligand will be shown in blue, bound to the protein. The protein residues included in the QM region will be shown in red. You can rotate the view using your mouse (left button) or translate the view with the mouse (right button). Double-click on any atom to reset that atom as the center for rotations. If you hover the mouse over any atom, information on the atom and the associated residue (name, number) will pop up. To close the pop-up window, click the “X” near the top left of the box.

• QM Region Summary: To the right of this section, you will find the QM Region Summary, which is populated after you validate the QM region contents. (The definition is auto-validated, and this region is auto-populated if you use the Suggest a QM Region button). In this section, you will find:
  • Status: Whether the contents of the definition box are properly validated. The possibilities are “Validated” or “Invalid.” If the contents are Invalid (there is a problem) or if they are valid but a warning was issued, the status will be shown in red, and you can ascertain more information by hovering the cursor over the question
    mark.
  • Number of link atoms: Number of hydrogen atoms that were added to address unfulfilled valencies that arose from using separated protein residues. These are required for the QM simulation.
  • Number of total QM atoms: Total atoms in the QM region, excluding the ligand itself. Based on the reference ligand.
  • Total QM charges: Total QM-determined charge in the QM region (ligand + other residues), based on the reference ligand.

4.10.3 Start Simulation

Once you are satisfied with your quantum region selections, you can start the simulation using the Start Simulation button at the bottom right of this part of the panel.

4.11 Simulation Status

This portion of the panel provides the status of the calculation once it has been started using the Start Simulation button. It also provides the ability to stop and restart a calculation that has been submitted.

• Stop: Stop a calculation that was previously submitted and is in progress. A stopped calculation is saved in the cloud storage associated with your account and can be restarted later, using the “Run” command.
• Run: Run a previously Stopped job. If the job has not previously been started using “Start Simulation,” then the Run button has the same effect as “Start Simulation,” i.e., it will begin the calculation.

• Light Grey: That segment of the calculation workflow has not been run.
• Dark Blue + Light Grey: That segment of the workflow is in progress and the dark blue portion of the segment reflects a progress bar.
• Dark Grey: That segment of the workflow has been completed.
• Red: That segment of the workflow is completed with errors.

Further down in this portion of the page, the status is provided at a more detailed level, showing the progress for every ligand mutation (every mutation represented by an edge in the mutation map. The names of the starting and ending ligands for each mutation are shown in the table, along with a four-segmented progress bar. The four segments for the progress bar represent Preparation/Equilibration/Production/Analysis. Each segment is colored as described above.

4.12 Results

Below the “Simulation Status” section, you will find the results of your calculation. The information in this section will be auto-populated when the simulations requested have been completed. The data is provided in four sections:

4.12.1 Result for Each Mutation

One result will appear in this table for every edge in the Mutation Map. The edges (lines) represent pairs of ligands between which free energy calculations will be carried out. The number of connections (edges) will typically be larger than the number of input ligands. The relative free energy changes for all the edges will be calculated. Performing more than the bare minimum number of changes required to connect the ligands helps provide more reliable net free energy difference values and associated errors in the Free Energy Results table (below). Note that information from the redundant pathways are not used in the Mutations results table, which provides energy values and errors determined from only the single mutation in each row.

All columns of the Result for Each Mutation table are sortable. Clicking on the header once will sort low-to-high. Clicking on the header a second time will sort high-to-low.

For each mutation that was run, the following will be reported:

• Mutation: Sequence number for the mutation calculation, starting from one.
• Start Ligand: The name of the starting ligand for the mutation. The name is either taken from the user input ligand file or, if not present in that file, assigned by the program as LNNN, where “NNN” is the counter for ligands without user-supplied names, which starts at 1. These names are auto-assigned, if necessary, for ligands supplied in the “All ligands (except reference)” input. For example, the first ligand read in without a name would be L001, and the tenth ligand read in without a name would be L010.
• End Ligand: The name of the ending ligand for the mutation. The name is obtained in the same fashion as for Start Ligand.
• ddG: The computed free energy of binding (ddG) (in kcal/mol) for the specified mutation between the Start and End Ligands. ddG is calculated as the difference [dG(complex) - dG(ligand)], which provides the net free energy of binding via the thermodynamic cycle relationship.
• dG(complex): The free energy (kcal/mol) calculated for the mutation between the Start and End ligands in the ligand/protein complex.
• dG(ligand): The free energy (kcal/mol) calculated for the mutation between the Start and End ligands calculated free in solution.

Below the Mutations table, an Export button is provided.

• Export table: Download the data in the Mutations table as a CSV-formatted file.
  
  

4.12.2 Accumulated free energy and convergence plots

If you select any row of the Status for Each Mutation table, then two graphs will appear as a pop-up.

• Accumulated free energy: This graph presents the accumulated ddG free energy differences versus lambda for the transformation selected in the Mutations table. The lambda values for the physical endpoints are always 0 and 1. Values are presented for the net ddG = [dG(complex) - dG(ligand)] at each value of lambda. Rolling the cursor over any point of the plot will show the (x,y) values for that point.
• Convergence with respect to sampling: This graph presents how the calculated net free energy for the mutation would vary if we used only a subset of the collected data, up to the maximum amount of data collected. From this curve, one can infer how well converged the results may be, and whether the sampling was sufficient. Rolling the cursor over any point of the plot will show the (x,y) values for that point.

In addition, 2D representations of the ligands corresponding to the two endpoints of the mutation are presented at the top of the pop-up window.

4.12.3 Free Energy Results

In this table, the net relative free energies for all the input ligands are presented. The net free energies are evaluated by averaging the energies for multiple paths (edges) that connect the ligand to the reference ligand. The Sampling Error represent the maximum error for any non-redundant closed loop of edges that includes the ligand.

All columns of the Free Energy Results table are sortable. Clicking on the header once will sort low-to-high. Clicking on the header a second time will sort high-to-low.

• Ligand: The name of the ligand. The name is either taken from the user input ligand file or, if not present in that file, assigned by the program as LNNN.
• ddG: The free energy of binding for the ligand (kcal/mol), relative to the reference ligand. The reference ligand is arbitrarily assigned a free energy of binding of 0.00. Relative free energy values ddG are calculated as ddG = [dG(complex) = dG(ligand)].
• Sampling Error: Sampling Error: Sampling error (kcal/mol) is estimated as the maximum error for any nonredundant closed loop that includes the ligand. If a mutation is not part of any closed loop defined by edges, then the estimated error is obtained from the error of the nearest ligand on a non-redundant closed loop. If there are no closed loops (i.e., a Task with a single mutation), the error is obtained from a combination of the statistical analysis carried out during the BAR processing for that mutation, plus the hysteresis for the calculation (energies calculated in ascending and descending direction for each lambda window). The sampling error is to be considered a lower bound.

Clicking on any row of the Free Energy Results table will show the 2D structure of the ligand for that row, on the right-hand side of the panel.

4.12.4 Result Download

At the bottom of the panel, you will find the Result Download button. You can use this button to download a ZIP format archive, named BATCH_NAME_raw_outputs.zip. (BATCH_NAME is the name you have given this batch calculation; For example, if your Batch was named FEP1, then the file would be named FEP1_raw_outputs.zip). Within this archive, you will find the following directories:

• ligands subdirectory: Contains a set of subdirectories with the input ligand structures, as provided by the user.
• receptor subdirectory: Contains the input protein receptor structure, as provided by the user.
• ligand_output subdirectory: Contains a CSV format file with the calculated free energies, one line per input ligand.
• mutations subdirectory: Contains several entities.
  • mutation_results.csv is the calculated relative free energies for each of the mutations (edges) defined by the mutations map, in CSV format.
  • specific mutations subdirectories: Each of these mutation subdirectories has a name of the format ligand1__ligand2, where ligand1 and ligand2 are the user-supplied (or program auto-generated) names. There is one such directory for every edge in the mutations map. Data within ligand1__ligand2 subdirectory is for the FEP calculation between nodes ligand1 and ligand2.
    
    Within each of these mutation (edge) ligand1__ligand2 subdirectories you will find
    
    bound (directory)
    unbound (directory)
    convergence_trajectory.csv
    lambda_trajetory.csv
    
    The two CSV files contain the data used to produce the convergence and accumulated free energy vs. lambda plots that are presented when you click on any of the mutations in the GUI.
    Within both the bound and unbound subdirectories will find a series of additional subdirectories, one for each lambda window used for that FEP calculation, each named
    
    windowNNN
    
    where N ranges from 001 to the number windows used. Within each of these subdirectories, you will find a PDB format file (“final.pdb”) that corresponds to the entire system (ligand/protein/waters/cofactors/ions) on the final step of the MD sampling trajectory used for that window. This file can be read into any standard molecular display program.
    
    In both the bound and unbound subdirectories, at the top level of that subdirectory, you will also find the file “initial.pdb”, which is the starting structure of the entire system (ligand/protein/waters/cofactors/ions) at the start of that window calculation.