Project description:Graph clustering is a fundamental problem in machine learning with numerous applications in data science. State-of-the-art approaches to the problem, Louvain and Leiden, aim at optimizing the modularity function. However, their greedy nature leads to fast convergence to sub-optimal solutions. Here, we design a new approach to graph clustering, Tel-Aviv University (TAU), that efficiently explores the solution space using a genetic algorithm. We benchmark TAU on synthetic and real data sets and show its superiority over previous methods both in terms of the modularity of the computed solution and its similarity to a ground-truth partition when such exists. TAU is available at https://github.com/GalGilad/TAU.
Project description:The leukemia-associated fusion protein MN1-TEL combines the transcription-activating domains of MN1 with the DNA-binding domain of the transcriptional repressor TEL. Quantitative photobleaching experiments revealed that ?20% of GFP-tagged MN1 and TEL is transiently immobilised, likely due to indirect or direct DNA binding, since transcription inhibition abolished immobilisation. Interestingly, ?50% of the MN1-TEL fusion protein was immobile with much longer binding times than unfused MN1 and TEL. MN1-TEL immobilisation was not observed when the TEL DNA-binding domain was disrupted, suggesting that MN1-TEL stably occupies TEL recognition sequences, preventing binding of factors required for proper transcription regulation, which may contribute to leukemogenesis.
