Skip to content Skip to navigation

Schrodinger


How to Run Schrödinger Software at Stanford

There are three main ways to run the Shrödinger software platform at Stanford:

1) On your local machine (Mac, PC or Linux):

2) Via an interactive GUI session on Sherlock (Stanford's awesome HPC! Thanks to the HPC crew!):

3) Via command-line scripts on Sherlock

Ultra Quick Guide to running Schrödinger in interactive mode on Sherlock 2.0 HPC

Step 1 - Request a Sherlock account:

http://www.sherlock.stanford.edu/docs/getting-started/prerequisites/#how-to-request-an-account

Step 2 - Login: 

http://www.sherlock.stanford.edu/docs/getting-started/connecting/

$ ssh -X username@login.sherlock.stanford.edu

Step 3 - Access a compute node:

$ sdev

Step 4 - Load the 'chemistry' meta-module and then the 'schrodinger' module: 

$ ml chemistry schrodinger

Step 5 - Run your job: 

http://www.sherlock.stanford.edu/docs/user-guide/running-jobs/

$ $SCHRODINGER/run myscript.py options

Note: $SCHRODINGER is the environment variable already setup for you, this parameter loads all the paths to find the software. Also note, the standard scripts detailed in the Schrodinger Script Center (see above) are already installed for you so there is not need to download.

Note: The "sdev" command is intended for quick jobs and will allocate one core and 4 GB of memory on one node for one hour. For more intensive jobs such as MD you will need to use SLURM to submit your job (see article above). You can request more resources for interactive jobs by following these instructions: http://www.sherlock.stanford.edu/docs/user-guide/running-jobs/.

Note: Login nodes are usually shared among many users and therefore must not be used to run computationally intensive tasks. Those should be submitted to the scheduler which will dispatch them on compute nodes.

Note: the -X option enables Maestro graphical GUI access.

Remember: connect to login.sherlock.stanford.edu

This gotcha catches me out all the time!! login directly to "sherlock.stanford.edu" will not work!

Sherlock uses load-balanced login nodes, so always remember to login to "login.sherlock.stanford.edu", also remember the -X option if you want to use the GUI via an X11 display:

$ ssh -X username@login.sherlock.stanford.edu

If you are connecting from a Mac or Linux it is pretty straightforward and you can use your built in terminal. However, connection from windows can be a little trickier and I would reccomend installing SecureCRT:

https://uit.stanford.edu/service/ess/scrt_sfx/securecrt

You can configure SecureCRT to access sherlock:

https://uit.stanford.edu/service/ess/scrt_sfx/add_sessions

Just remember to configure the session to connect to "login.sherlock.stanford.edu"!!

Running the Maestro GUI on Sherlock

Here is a cool way of running the Maestro GUI on Sherlock:

$ ssh -X username@login.sherlock.stanford.edu

$ sdev

$ ml chemistry schrodinger

$ maestro &

Note: the -X option passes the display to your local X11 display. If you are on a Mac or Linux system you should be good to go, on Windows you will need to install a thirdparty X11 server such as Xming X Server.

Note: This is useful for more intensive jobs that take too long on your laptop or desktop, but please be aware that this method will not get you complete access to the compute power of Sherlock (as your jobs will be limited). To access full compute potential you will need to submit a commad-line job via SLURM (see below).

Note: If you are using Linux or Mac you can use a simple -X option on your SSH command (as above). On windows you will need to configure your client, for example SecureCRT, to forward the display to your local X11 display (Properties-->Port Forwarding-->Remote/X11--> Check Forward X11 Packets).

Submitting jobs with SLURM

Step 1: Write your submit.sh script:

#submit.sh
#!/bin/bash
#
#SBATCH --job-name=test
#
#SBATCH --time=10:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=2G
#
ml chemistry schrodinger
$SCHRODINGER
srun myscript.py

Step 2: Submit the job to the scheduler:

$ sbatch submit.sh

Step 3: Check your job:

$ squeue -u $USER

Step 4: Check the output file, where ### is the JOBID number from Step 3:

$ cat slurm-###.out

Schrödinger Script Center

The Schrödinger Script Center is the best place to find the Schrödinger command line scripts to get you up and running fast, these scripts are already installed on Sherlock for you:

https://www.schrodinger.com/scriptcenter

Run a command-line script like this:

$ $SCHRODINGER/run fragment_join.py options

...where "$SCHRODINGER" is the enviroment variable (i.e. link to the software and paths etc) already setup for you and "options" are flags relevant for the program you are running.

Further information, and "options" flags, may be obtained for scripts by executing them with the "-help" flag, for example:

$ $SCHRODINGER/run fragment_join.py -help

Here is a summary of job option flags.

Biologics Workshop Feedback and Training Materials

Thanks to everyone that attended the first ChEM-H/MSKC sponsored training workshop. We would appreciate your feedback using the form below, thank you.

Here are the training documents:

Schrödinger@stanford email List

Join the Schrodinger@stanford email list here:

Schrödinger Biologics Suite - Protein Surface and Interface Analysis

There are some excellent professionally produced tutorials below, but here is my hack attempt at getting the biologists amongts us up and running fast.

This quick tutorial focusses on protein surface and interface analysis.

Installation Guide

Here is a quick installation guide. This is obviously a PC installation, but Mac installation is basically the same (swap the .zip file for .dmg package etc etc).

Feel free to email me with any simple problems or questions. Bugs and super technical questions should go to Schrödinger tech support.

Site License Announcement

Great news! Stanford finally has a campus wide license for Schrödinger software including the Biologics Suite, Small-Molecule Drug Discovery Suite and Materials Science Suite.

Thank you to Grace Baysinger from the Stanford Library for her leadership on this project and also thank you to Addis O'Connor for implementation on the Sherlock HPC cluster. Thanks also to the Vice Provost for Teaching and Learning for supporting the project and ChEM-H for providing follow up training.

Full details, including download and license instructions, can be found on the links below.

Running Desmond on a GPU

We have received reports of successful executation of Desmond on the GPU of a local linux machine. The user was able to run a 100 nanosecond simulation in 41 hours. The system contained a protein in a fully hydrated bilayer with a total of 89,000 atoms. The configuration of the system is as follows:

  • Linux Unbuntu 16.04 (64-bit)
  • Nividia Quadro M6000 (12Gb VRAM), with Nvidia driver 384.98

This graphics card has over 3000 cores, so it will be interesting to compare how these runtimes compare with execution on an HPC such as Sherlock.

Below is a link with tips for running molecular simulation on your local GPU.

Thanks to Edward J. Bertaccini, MD (Dept. of Anesthesiology) for sharing these specs with us.

Protein Ligand Database (PLDB)

The Protein Ligand Database (PLDB) is part of the Schrödinger Suite available at Stanford.

The PLDB is a fully annotated and curated version of the PDB that you can use to run querys against. The types things you can query include:

  • Density data: fo-fc and 2fo-fc electron density maps.
  • RCSB metadata (deposition date, resolution, etc.).
  • The capability to search for interactions between protein and ligands​, incuding Hydrogen bonds, Pi-Pi interactions (face to face, face to edge), Halogen bonds, Salt bridges and Metal-ligand/protein bonds.
  • Powerful geometric search criteria, including distance, angle, dihedral, centroids, vectors and planes.
  • Predefined Kinase features, including gatekeeper residue, catalytic residues, C-Helix, DFG motif, hinge regions, activation loops, P-Loops and HRD.
  • And much more

To use PLDB:

In order to use the PLDB, Schrodinger staff must first create an account for a user.  Requests for new accounts should be sent to graceb@stanford.edu.
 
Users who want to use the PLDB should also review the Schrödinger EULA. This is the same EULA that you agree to when you downloaded the Schrödinger Suite.   http://www.schrodinger.com/salesagreements/

To run PLDB:

  • Select the PLDB Search... option from the Structure Analysis section of Maestro's Tasks menu.
  • ...or by searching for PLDB using the search bar at the top-right of Maestro.
  • This will present you with a login dialog where you will enter the following:
  • Server URL: https://pldb.onschrodinger.com
  • User Name: your Stanford email address, e.g. joeblogs@stanford.edu 
  • Password: stanford_pldb

You can find a lot more details, and some examples, in the MSKC Blog at the link below.

Super Quick Method for Running MD Simulations via the GUI on Sherlock

Here is a quick way of running Molecular Dynamics simulations (MD) via the GUI on Sherlock. You can, of course, run DESMOND on you local machine, or via scripts using SLURM, but by far the easiest way to get up and running quick is by using the GUI in interactive mode on SHERLOCK. A typical MD simulation will take ~48 hours (or much longer), so you definately need some high-end computational resources.

Warning: The GUI will be slow and unresponsive on slow networks or via Wi-Fi. Also, remember you will need to be on a Stanford networks to get the license server to work correctly.

1) Run the GUI in interactive mode on Sherlock:

​On Linux/Mac you can use the built in terminal. On Windows SecureCRT works well. See above.

$ ssh -X username@login.sherlock.stanford.edu

$ sdev

$ ml chemistry schrodinger

$ maestro &

2) Preprocess your stucture using the Protein Preparation Wizard.

3) Search for "Desmond" in the search tool at the top right under the "Tasks" menu.

Note: the -X option passes the display to your local X11 display. If you are on a Mac or Linux system you should be good to go, on Windows you will need to install a thirdparty X11 server such as Xming X Server.

Note: If you are using Linux or Mac you can use a simple -X option on your SSH command (as above). On windows you will need to configure your client, for example SecureCRT, to forward the display to your local X11 display (Properties-->Port Forwarding-->Remote/X11--> Check Forward X11 Packets).