Matcha

Motility Analysis of T-Cell Histories in Activation

Overview

Despite advances in microscopy, imaging techniques only capture a very small spatial and temporal fraction of T-cell activity within a tissue. For this reason, we have pursued the development of a virtual T-cell model. Matcha's objective is to design virtual T cells that move like biological T cells. The virtual T cells will match the speed and turning angle distributions of biological cells. In addition, Matcha will also attempt to match the volume patrolled by the biological cells and the dependence on speed and turning angle on the previous speed.

Prerequisites

1. A Fortran 2018 compiler with multi-image/coarray support (e.g. gfortran + OpenCoarrays)
2. The Fortran Package Manager

Downloading, Building, and Testing

GNU Compiler Collection (GCC) gfortran Compiler

Single-image (serial) execution

With gfortran installed, download, build, run and test Matcha in a single image by entering the following commands in a bash-like shell:

git clone https://github.com/rouson/matcha
cd matcha
fpm test

Multi-image (parallel) execution

With gfortran and OpenCoarrays installed, build, test, and run Matcha in multiple images by entering the following commands in a bash-like shell:

fpm test --compiler caf --runner "cafrun -n 2"
fpm run --compiler caf --runner "cafrun -n 2"

Change '2' above to the number of images that you would like to launch in parallel.

Intel OneAPI ifx Compiler

Multi-image (parallel) execution

The unit test suite uses small grids to minimize runtime, which limits the scalability of the test suite only. Hence, the test suite is not designed for execution in a large number of images.

All tests pass when running the test suite in 1-4 images:

export FOR_COARRAY_NUM_IMAGES=4
fpm test --compiler ifx --profile release --flag "-heap-arrays -coarray"

Note this requires a working install of Intel MPI.

Single-image (serial) execution

Same as multi-image execution except FOR_COARRAY_NUM_IMAGES=1. Also requires a working install of Intel MPI.

Automatic GPU-based acceleration: Experimental

The following command is listed here for reference while we investigate what appear to be platform-specific compiler issues causing a crashe when automatic GPU-based acceleration is enabled with the required coarray features:

fpm test \
  --compiler ifx \
  --profile release \
  --flag "-heap-arrays -coarray -fopenmp-target-do-concurrent -qopenmp -fopenmp-targets=spir64"

Documentation

Please visit the Matcha Github Pages site to see HTML Documentation generated with ford.