Modeling complex systems
Biosystems Analytics has consulted with many scientists and businesses helping them conceive and prototype complex systems models ranging from development genetics, ecological fish dynamics, bacterial colonies to supply chains and stockmarkets. To this end, we use a multifacted approach to modeling including agent-based, deterministic and stochastic strategies, as well as hybrid approaches combining all of the above. Projects we have consulted on, prototyped, or fully implemented include:
- Systems biology: models of transcription factor network evolution that incorporate biological realistic details drawn from experimental data, yet are computationally tractable.
- Pharmacology: models of individual variability in drug response systems using agent-based modeling
- Ecology: agent-based model to explore scaling relationships between canopy size, tree size in forests.
- Computational neuroscience spatial agent-based model to model the evolution of neural networks to test hypotheses about how hierarchical structures emerge.
- Theoretical biology Extended AlChemy (ALgorithmic CHEMistry), an artificial abstraction of chemistry to include a spatial dimension to investigate the evolution of cellular membranes.
Complex systems modeling software including Swarm
We were also active in developing Swarm (swarm.org), the pioneering open-source agent-based modelling toolkit from the Santa Fe Institute (Dr. Lancaster was one of its two primary software developers from 1997-2000) as well as consulting with scientists to prototype and building models using the toolkit (see above). We also have experience in other agent-based tools and technologies and also worked in the UC Berkeley based team to extend the differential equation-based modeling tool, Berkeley Madonna to Java.
PLAAC: searching sequences for structure
PLAAC (Prion-Like Amino Acid Composition) searches protein sequences to identify candidate prion subsequences using a hidden-Markov model (HMM) algorithm. The underlying code for the PLAAC algorithm, as well as the web-frontend is available under the open-source MIT license via GitHub: http://github.com/whitehead/plaac The main website for PLAAC is at: plaac.wi.mit.edu.
PyPop: probing genotype data for signatures of selection
Python for Population Genomics (PyPop) is an open-source framework for large-scale population genomics, implemented to probe genotype data for signatures of natural selection. It has been used extensively in the population and statistical genetics community in over peer-reviewed 150 publications. Written in Python it generates standardized XML output format. Binaries are available for Linux and Windows, and full source released under the GNU GPL (http://github.com/alexlancaster/pypop), available at: pypop.org
COSMOS: building scientific workflows
Member of the team that developed the open-source library COSMOS (cosmos.hms.harvard.edu): a Python library to manage large-scale workflows that allows formal description of pipelines and partitioning of jobs developed for (but not restricted to) running next-generation sequencing (NGS) workflows. It includes a user-interface for tracking the progress of jobs, abstraction of the job queuing system (to allow interface to multiple queuing systems) and fine-grained control over the workflow. COSMOS runs on cloud-based services such as Amazon Web Services and Google Cloud, as well as traditional high-performance computing (HPC) clusters. It has recently been open-sourced under the GNU GPL license: http://github.com/LPM-HMS/COSMOS-2.0 along with a genome analysis workflow, GenomeKey: https://github.com/LPM-HMS/GenomeKey (under the MIT license).