Differences

This shows you the differences between two versions of the page.

--- public:ba-themen [2024/08/27 07:26] – spohr
+++ public:ba-themen [2025/12/09 13:53] (current) – Added another POTTR related topic schulte
@@ Line 13: / Line 13: @@
   * Integrate/Implement CP ILP Formulations as a stand-alone application that processes MIDI or as a plugin for the scorewrite software MuseScore (Tanaka) Contact: Philipp
   * Paper Dominik Heider https://www.sciencedirect.com/science/article/pii/S0010482524002695 do that with LOBICO (categorical classification, not binary)
-  * Maximizing diversity for anticlustering (while keeping optimal dispersion). Keywords: ILP, R, C++, maybe preprocessing, maybe optimization with constraint programming. Contact: Khoa
+  * Changing the calculation of a drug response metric for a pharmacodynamic dose-time-response model. Contact: Khoa
-  * "Continuous logical" formulas for regression to predict drug responses. Keywords: ILP, Python. Contact: Khoa
+  * Removing tautomers and mesomers in generated chemical structures. Contact: Khoa
-  * 3D-time-dose-response surface fitting and extrapolation, with focus on different time-response models. Keywords: Integral, Gompertz-Laird/exponential functions. Contact: My Ky
+  * Clustering of tumor trees with POTTR / based on POTTR. Contact: Sara
-   * Implementierung und Vergleichen von Metaheuristiken für Maximum Diversity bei Anticlustering [[https://doi.org/10.1016/j.ejor.2022.02.003]]. Contact: Nguyen Khoa Tran
+  * Find maximal common trajectories in conflict graph using heuristics for max independent set. Contact: Sara
 === MSc level ===
+  * Development and implementation of a multi-grid Alignment Algorithm leveraging structure-adapted banded alignments. Contact: Martin
+  * Test and develop approximation algorithm for finding maximum / maximal common trajectories, probably combine it with an optimization after finding approx. solution. Contact: Sara
   * Vehicle routing problem for leaf optimization. Get Martin's code to run. Produce better pictures for Daan. Make modifications to speed up. Check if the model is good enough, adapt, ... Contact: Gunnar
   * Data from Olga. compare molecular graphs. Contact: Gunnar
   * Further stuff Nan (new stuff on image analysis)
+=== Projektarbeit ===
+  * Extend PHI (https://doi.org/10.1101/2024.10.27.620212) to work with diploid organisms
+  * Histo-Slide stitching via partial differential equations - Image-Puzzling for tissue samples. Contact: Martin
 === Other topics / No fixed level ===
+  * Using Evo2 to improve Metagenomics Classification, specifically for Unmapped Reads, Contact: Philipp
+  * Benchmarking and Testing of Clustering-based, Manual, and ML-based Cell Type Prediction in High-Plex Imaging Data (Bioinformatic M.Sc. project)
+This project aims to evaluate different approaches for cell type prediction in high-plex imaging data, including clustering-based methods, manual annotation, and machine learning (ML) models. The study will compare the accuracy, reproducibility, and computational efficiency of these techniques across diverse datasets. The findings will provide insights into the optimal strategies for cell identification in spatial omics studies, supporting more robust downstream biological interpretations.
+  * Application and Optimisation of Automated Cell Type Prediction Tools for Distinct Hematologic and Solid Malignancies to Identify Distinct Changes in Patient Subgroups (Quantitative Biology M.Sc. project)
+This project focuses on optimizing automated cell type prediction tools for specific hematologic and solid malignancies in multiplexed immunofluorescence imaging. By applying advanced computational methods taking into account the known composition and characterisation of the disease entity, the study will identify unique cellular alterations within specific patient subgroups, potentially revealing novel biomarkers or therapeutic targets. The work involves testing and refining prediction models to improve their accuracy in detecting disease-specific cellular patterns, contributing to precision oncology research. While this project focuses on the computational side of multiplexed immunofluorescence, interested students can also get hands-on experience in data generation and wet-lab techniques.
   * Efficient heuristic or exact algorithms for circular sequence comparison in sub-quadratic time: Aligning bacterial genomes vs one another or reads of a circular genome vs a circular genome
   * SWGTS-Expansions: 1.) Add encrypted package transfer and key sharing for increased throughput/speed, 2.) Alternative Filtering Methods e.g. k-mer based filtering 3.) Metadata Managment System 4.) Division of long reads into segments ;  Skills: Docker-Compose, REST, Encryption, Client-Server Architecture
@@ Line 32: / Line 43: @@
   * Spa-Typing and Flight Data: Collect Flight Data and Samples from Spa Database and look for Correlations (Check if a high volume of flights correlates to similar detected Spa-Types / Develop a framework to check correlation between flight volume and genetic information), Build on Ninas work, Contact: Philipp
   * Backport Sven's 0-Edge-CE-Heuristic into Yoshiko Main (and thus Cytoscape App), Contact: Sven, (Philipp?)
-  * Projektarbeit, very unclear outcome: https://newatlas.com/massive-attack-dna-spray-paint/56881/ Decode Sequence Massive Attack