Skip to Content

 Thumbnail Image

Biosystemix Ltd.


Pioneering Data-Driven Biomedical Discovery
for Personalized Medicine and Mechanistic Inference


Biosystemix provides cutting-edge solutions in advanced computational analysis and predictive model development for personalized medicine, therapeutic discovery, and disease genomics.

Biosystemix’s flagship product, the READ diagnostic test (Risk Evaluation for AHCT Donors), was developed within the S2K research partnership with Claude Perreault MD, the University of Montreal, and Genome Quebec. READ will transform the clinical practice of bone marrow transplantation by eliminating donors who cause the debilitating tissue rejection called GVHD (Graft Vs. Host Disease).

Our technologies cover the full range of computational methods and domain expertise required for innovative biomedical data analysis and personalized medicine solutions:

  • Statistical data assessment, exploratory analysis, and visualization
  • Advanced data mining and predictive outcome modeling
  • Data-driven, molecular signaling network inference and pathway discovery

To help our customers meet their objectives, our service solutions are currently being applied in personalized medicine and disease genomics discovery partnerships:


The higher definition achieved with Biosystemix expertise supports predictors and models that show superior performance, and often enables solutions where conventional approaches may come up empty-handed.  Given today’s biomedical investments into data-intensive molecular and clinical R&D for diagnostic/prognostic marker discovery, pathway inference and compound efficacy/toxicity evaluation, Biosystemix advanced analysis is helping our partners obtain more value from these efforts.


Biosystemix reports personalized medicine advances


Prediction of Graft-Versus-Host Disease (GVHD) in Humans

Biosystemix scientists, in a discovery partnership with Dr. Claude Perreault (University of Montreal), report significant progress in solving the difficult problem of predicting whether blood stem cell donors will induce GVHD in recipients.  Detailed methods and findings were reported recently in the journal Public Library of Science, Medicin:  Baron C, Somogyi R, Greller LD, Rineau V, Wilkinson P, et al. (2007) Prediction of graft-versus host disease in humans by donor gene-expression profiling. PLoS Med 4(1): e23. doi:10.1371/journal.pmed.0040023

Graft-versus-host disease (GVHD) results from recognition of host antigens by donor T cells following allogeneic hematopoietic cell transplantion (AHCT). Notably, histoincompatibility between donor and recipient is necessary but not sufficient to elicit GVHD. Therefore, we tested the hypothesis that some donors may be “stronger alloresponders” than others, and consequently more likely to elicit GVHD.

To this end, we measured the gene expression profiles of CD4 and CD8 T cells from 50 AHCT donors with microarrays. We report that pre-AHCT gene expression profiling segregates donors whose recipient suffered from GVHD or not. Using quantitative PCR, established statistical tests, Biosystemix Predictive Interaction Analysis (PIA) and analysis of multiple independent training-test datasets, we found that for chronic GVHD the “dangerous donor” trait (GVHD+ recipient) is under polygenic control and is shaped by the activity of genes that regulate TGF-β signaling and cell proliferation. We also performed gene expression profiling in T cells harvested from 40 AHCT recipients on day 365 post-AHCT. The donor gene profile defined on day 0 showed exceedingly strong correlation with that of recipient CD4 and CD8 T cells harvested one year post-AHCT.

These findings strongly suggest that the donor gene expression profile has a dominant influence on the occurrence of GVHD in the recipient. Donor gene profiles linked with GVHD were probably imprinted at the hematopoietic stem cell level, because they persisted long-term in the recipient. The ability to discriminate strong and weak alloresponders using gene expression profiling could pave the way to personalized transplantation medicine.


Predicting drug response in multiple sclerosis

Biosystemix scientists contribute to solving the difficult problem predicting a patient’s drug response from molecular profiling of heterogeneous patient samples in a complex progressive autoimmune disease. Detailed methods and findings were reported recently in the journal Public Library of Science, Biology  (Baranzini SE, Mousavi P, Rio R, Caillier SJ, Stillman A, Villoslada P, Wyatt MM, Comabella M, Greller LD, Somogyi R, Montalban X, Oksenberg JR (2004) Transcription-based prediction of response to IFNb using supervised computational methods.  PLoS Biol 3(1): e2). In this challenging MS study, scientists maintained the highest quality standards regarding precision gene expression measurements, advanced data mining, predictive modeling, and in-depth statistical validation (a robust combination of IBISTM methodology and extensive re-sampling statistics.

In a perspectives commentary in PLoS Medicine (Kaminski N, Achiron A (2005) Can blood gene expression predict which patients with multiple sclerosis will respond to interferon? PLoS Med 2(2): e33 ), researchers Kaminski and Achiron captured the essence of the MS-3d IBIS study:  "The importance of Baranzini and colleagues’ study lies not in its mechanistic insights, but in its clinical relevance. The careful design of the experiment, the use of reproducible real-time PCR instead of microarrays, the meticulous analysis, and the previous observations support the notion that PBMCs [peripheral blood mononuclear cells] express clinically relevant gene expression signatures in MS [multiple sclerosis] and probably in other organ-confined diseases.”



Biosystemix was founded by Dr. Roland Somogyi and Dr. Larry D. Greller to serve the rapidly growing demand for data-driven discovery for enabling today’s advanced medical solutions.  We provides the technologies, data analysis services, and partnering solutions to build the bridge between molecular and clinical study data and effective personalized medicine application models.

 Text Box:      Successful personalized medicine and mechanistic discovery programs ultimately depend on understanding the data and deriving meaningful predictive models

Our efficient and direct, data-driven approaches rely on advanced methods from statistics, signal processing, machine learning, pattern recognition, data mining, and mathematical modeling.  Our work is carefully orchestrated with biomedical domain expertise to guide the definition and search for effective applications.  Biosystemix currently provides its expertise in the form of consulting, analysis services, discovery partnerships, and reports focusing on complex predictive models for customer applications.  Together with our partners we are developing novel solutions that combine biomarkers and therapeutics into personalized medicine application solutions.


Biosystemix has a track record of first-in-the-field, high-quality publications in computational biomedical applications:

  • First predictive clinical model of GVHD (graft-versus-host disease) from blood cell gene expression:  Baron C, Somogyi R, Greller LD, Rineau V, Wilkinson P, Cho CR, Cameron MJ, Kelvin DJ, Chagnon P, Roy DC, Busque L, Sékaly R-P, Perreault C (2007) Prediction of graft-versus-host disease in humans by donor gene expression profiling. PLoS Med 4(1): e23
  • First detailed model explaining why pulsatile parathyroid hormone release leads to increased bone formation:  Potter LK, Greller LD, Cho CR, Nuttall ME, Stroup GB, Suva LJ, Tobin, FL (2005) Response to continuous and pulsatile PTH dosing:  A mathematical model for parathyroid hormone receptor kinetics, Bone 37, 159-169 (2005)
  • First fully cross-validated 3-d Bayesian personalized medicine model for predicting drug response:  Baranzini SE, Mousavi P, Rio R, Caillier SJ, Stillman A, Villoslada P, Wyatt MM, Comabella M, Greller LD, Somogyi R, Montalban X, Oksenberg JR (2004) Transcription-based prediction of response to IFN using supervised computational methods.  PLoS Biol 3(1): e2
  • First data-driven, reverse-engineered model of gene interaction networks derived from measured, high-fidelity gene expression data: D'haeseleer P, Wen X, Fuhrman S, and Somogyi R (1999) Linear Modeling of mRNA Expression Levels During CNS Development and Injury. Pacific Symposium on Biocomputing 4:41-52
  • First gene network reverse engineering algorithm within the framework of discrete Boolean network models:  Liang S, Fuhrman S, Somogyi R (1998) REVEAL, A general reverse engineering algorithm for inference of genetic network architectures. Pacific Symposium on Biocomputing 3:18-29
  • First cluster and pathway analysis study of large-scale, high-fidelity gene expression time series data:  Wen X, Fuhrman S, Michaels GS, Carr DB, Smith S, Barker JL, Somogyi R (1998) Large-Scale Temporal Gene Expression Mapping of CNS Development. Proc Natl Acad Sci USA 95:334-339
  • First nonlinear stability analysis study explaining the nature of hormonal tuning of recorded liver Ca2+ oscillation data:
    Somogyi R, Stucki JW (1991) Hormone Induced Calcium Oscillations in Liver Cells Can Be Explained by a Simple One Pool Model. J Biol Chem 266:11068-11077


Biosystemix uses a unique mixture of algorithms and workflows, covering novel internally developed and more broadly established methods, which are then integrated into customized, project-specific solutions for each customer:

IBISTM (Integrated Bayesian Inference System – compute-intensive cross-validation for multivariate, multiclass Bayesian inference of outcome probabilities), LDA and QDA-based,  univariate and multivariate PIA (Predictive Interaction Analysis – inferring interactions through outcome discrimination and prediction), pair-wise gene-gene (variable-variable), combinations predictive of outcome, prioritized according to comprehensive statistical scoring, CPIA (Competitive Predictive Interaction Analysis), SPIA (Synergistic Predictive Interaction Analysis); TEA (Theme Enhancement Analysis -  linking data-supported biological functional themes to outcome discrimination and prediction), statistically-supported enhancements of informative gene groups; PI2 (Pathway Interaction Inference) through combined PIA and TEA, inference of competitive and synergistic pathway interactions, associations of pathway interactions with clinical and biological outcomes; Gene Network Reverse Engineering, cofluctuation analysis (associations across time, or condition, or assay, etc), continuous analysis, discrete analysis, linear and nonlinear analysis, multivariate analysis, cluster analysis, graph analysis, clique (identity cluster) extraction, multi-input graphs; ANOVA, F-test, multi-class tests, T-test, 2- class tests; MANOVA (multivariate ANOVA), 2- class tests, multi-class tests; Chip and class similarity analysis, Pearson correlation, Euclidean, other similarity measures as needed, Concordance, means of class-distances, distances of class-means; Discriminant Analysis, LDA (linear discriminant analysis), QDA (quadratic discriminant analysis), 2-class analysis, multi-class analysis, univariate, multivariate.


Biosystemix, Ltd.  ©  2007