Features

High Throughput

ChemFlow was tuned to optimally use the high performance and throughput of available computing resources, following architectural standards of SLURM and PBS queueing systems. The job distribution was designed to minimize I/O and maximize throughput. Simplified configuration files allow them to adapt this policy to the available topology.

Note

We’ll soon add support to Sun Grid Engine.

Fault tolerance and Checkpointing

A big concert when dealing with high throughput data is to be able to diagnose and resume after unsuccessful execution. ChemFlow detects and report failed jobs, and resubmit them all.

Warning

One should always investigate the reasons of failure with proper care.

Analysis and Reporting

The major benefit from standardization is to facilitate analysis. Currently ChemFlow ships with protocols perform, analyses and report for some common scenarios for computational chemistry.

1. Docking and virtual screening of compounds including prospection, validation and method comparison.
2. MD simulations, contains protein stability (temperature, solvent, solvent mixture).
3. Any two-molecule system contains “rescoring” and binding free energy estimation.

Note

Analysis are mostly implemented with Bourne-again Shell (BASH) while the reporting tools are implemented as interactive Python Notebooks.

Data curation and reproducibility

Data curation demand for industrial applications require compliance with ISO/IEC 17025, the standard for which laboratories must hold accreditation in order to be deemed technically competent. The ChemFlow standard is a readily accessible and familiar specifications useful for data curation on drug discovery and molecular simulation projects.