The aim of this personal project is to apply and improve skills I learned in my Web Technologies course and create a cool extension of my resume that uniquely identifies me.

I also took it as an opportunity to improve myself and work on self-motivation, attention to details and creatvity by designing everything from scratch.

Here we present the Bioinformatics Master students’ carbon consumption during their apprenticeship.
We gathered all jobs carried out by the students for a month and determined their environmental impact using the Green Algorithms tool.
We studied how these results break down through the bioinformatics domains under scrutiny.

GreenBIM Poster

Comparison of in silico spectrums with a reference MGF file.

Set an ion intensity treshold for a set of spectrum and count the nunber of hits. Hits above the treshold will have a doubled score.

Add an acetylation mass to all candidate peptids and compute ions masses.

Return for each tested candidate peptid the number of hits, ranked by quality.

Subject: Create a HTML page and a PDF report from a XML file.

Write a XSLT transformation file to create an HTML page, according to a template, from a XSL data file.

Write a XSL-FO transformation file to create a PDF report from the same data file.

Subject: Implementation and modification of the LSL-LAE (Living Systems Lab Auto Encoder) model.

Retrieve the LSL-AE code and run it on a single-cell RNA-seq dataset.
Use code from the JAE (Joint Auto Encoder) to integrate SVD (Singular Value Decomposition) in the LSL-AE model.
Benchmarking of the AE and VAE (Variational Auto Encoder) models.

Pipeline for epigenomics data analysis (ChIP-seq experiment on E. Coli).

FASTQC - Quality control of the reads and statistics.
Bowtie - Mapping.
Picard MarkDuplicates - Estimation of the number of duplicated reads.
Deeptools - ChIP quality control.
IGV - Data visualisation.
MACS2 - Peak calling.
RSAT - Motif Discovery.
HOMER - Peaks annotation.

Subject: From a public bacterial genome's sequence, model this sequence using a Markov Model and use a Hidden Markov Model to predict coding regions.

Find a Markov Model fitted to the data using (1) Likelihood ratio test and (2) AIC and BIC criteria.
Detect coding and non coding regions:
(1) Learn a M1-M0 model.
(2) Use Viterbi algorithm to predict the most probable hidden state sequence.
(3) Repeat for a M1-M1 and a SM1-M1 model.
(4) Compare the results with a gene prediction tool (GeneMarkS).

Assign genes annotations for a given species : Gene Ontology terms, Reactome and KEGG pathways, and Protein domains .

Perform enrichment analysis using 2 approaches : Gene Set Enrichment Analysis (GSEA) and Over Representation Analysis.

Enrichment Analyses were performed using the ClusterProfiler R package and were manually re-implemented in R as well. We used Pathview to create figures of the pathways and BiomartR to convert gene ids. All figures were created using ggplot2.

Analysis of publicly available 3D complexes of human AChE co-cristallised with denepezil and galanthamine (PyMOL).
Docking of donepezil and donecopride in the AChE model with Vina Autodock.

Collection of scripts to aggregate gene annotations in an interactive tabulated file for a given species. We used BioPython, the REST API and the NCBI E-utilities API to retrieve annotations from several public databases (NCBI, Ensembl, Uniprot, Prosite, Reactome, KEGG, Gene Ontology). The interactive table was created with DataTables.

Development of a User Interface with the tkinter Python library.

Subject: Create a neural network of subthalamic nucleus neurons in Rattus norvegicus and study this network's behaviour by simulation.

Creation of 4 independent subthalamic nucleus neurons.
Creation of dendritic trees
Construction of a neuronal connexion.
Simulation and observation of the somatic membrane potential of different neurons. Analysis of the differences between the signals.

Analysis of the signals characteristics in the switch from a right to a wrong key.

Sounds were gathered with audacity and saved in R as Wave objects (tuneR package).
Several parameters were observed : signal amplitude during time, harmonics and spectral analysis, autocorrelation of time-based variables.
Finally, I observed what happened during the transition from a right to a wrong key.

Subject: As an artists agent, you have to create a database to keep track of all your clients.

We decided (in pair) to create a database of French rappers and regroupep informations such as the type of contracts, our artists commitmments, their salary, royalties, their agenda as well as the time, date, and location of their concerts, showcases and other events.

The project was carried out in 4 steps : a - Database modelling; b - Creation and filling of the tables; c - Commom operations and mandatory requests; d - Other requests database specific.

Tablut is a traditional Viking Game. One player controls the Shields and the other controls the Swords. The Swords player must try to capture the Chief Shield, while the Shields player must clear a safe path for the Chief Shield to escape off the board. Pieces can move any number of spaces in a straight line, and are captured when bracketed on two opposite sides by opponent pieces.

Play tablut online

Microsporidian genome sequencing highlighted the presence of transposable elements with a considerable diversity. MicroAnnot integrates a module based on the exploitation of TransposonPSI and RepeatMasker that allows the identification of genes. We integrated InterProScan to MicroAnnot to perform the functional analysis of these genes. The use of a specific microsporidian transposable elements database allowed an extensive characterization of the genes within the genome of Tubulinosema ratisbonensis as well as their functional annotation.