Danesh M.
Data Scientist

United Kingdom-London

8+ years experience

Interested in this profile?




Center for Computational Biology

Jan 2018

Group Leader

* Established and led a growing team of data analysts, specialized in big genomic data analysis.
* Developed and optimized a machine learning framework, based on ensemble machine learning
models, to accurately predict bacterial growth from genomic data.
* Designed and delivered courses on machine learning, data analysis and bioinformatics at the University of Birmingham.

Wellcome Trust Sanger Institute

Jan 2014 - Jan 2018

Senior Data Scientist

* Machine learning approaches to predicting antibiotic resistance (PanPred):
Developed a generic machine learning framework in Python to accurately predict antibiotic resistance from genetic and epidemiological input features. The model invoked logistic regression, random decision forests, gradient boosted decision trees and deep learning techniques. Gradient boosted decision trees consistently outperformed alternative models. The best model attained an overall accuracy of 0.92.
* Data analysis of large-scale genomic data: Employed a wide range of statistical models, e.g.
regression and Bayesian models, in R and Python to analyze 7 large-scale genomic datasets.
* Prepared reports, visualized and presented results at meetings and conferences.


Jan 2011 - Jan 2014

Research Assistant

Thesis project: My dissertation work examined the impact of recombination (shuffle crossovers) on adaptation (optimization), using deterministic and stochastic mathematical models.
* Recombination on HIV fitness landscape (RecombinationHIV): Successfully established a novel stochastic evolutionary algorithm in Java to simulate the evolution of HIV genomic sequences over time. The aim was to evaluate the impact of shuffle crossovers on the rate of optimization on a realistic fitness landscape.
* Evolution of natural competence (ClonalComplex): Developed two novel mathematical and computer simulation frameworks in Java and Mathematica. The models were based on continuous- time Kolmogorov forward equations. The models effectively simulate complex dynamics of large populations.
* Evolution of recombination on rugged fitness landscapes
(RecombinationRuggedLandscape): Designed and implemented models in Java and Mathematica
to address an important question in the evolutionary algorithms, i.e. the impact of shuffle cross- overs on the rate of optimum finding on complex fitness landscapes. The framework allowed to confirm that, irrespective of the fitness landscape structure, shuffle crossover accelerates optimization.

Department of Bioinformatics – George August University – Göttingen

Jan 2010 - Jan 2010

Research Assistant/ Data Scientist in Bioinformatics

• Implemented a novel machine learning algorithm (Hidden Markov Model) to find new patterns in Human genome in Java. The model was based on conditional random fields.


Jan 2010 - Jan 2010


2018 Prediction of antibiotic resistance in Escherichia coli from large-scale pan-genome data.
2015 Recombination accelerates adaptation on a large-scale empirical fitness landscape in HIV.
2013 The effect of bacterial recombination on adaptation on fitness landscapes with limited peak accessibility.


Jan 2010 - Jan 2010

Honors and Awards

2015 Silver medal of the ETH Zurich for outstanding thesis work
2014 Prospective Researchers Fellowship, Swiss National Science Foundation
2009 Stipend of the Excellence Foundation for the Promotion of the Max Planck Society


English language quiz B2
Python 3 quiz beginner level

Newest members that made a profile