MatrixMotif

0519622 - MBÚ 2020 RIV CZ eng L4 - Software
Vališ, Karel - Novák, Petr
MatrixMotif.
Internal code: MatrixMotif ; 2019
Technical parameters: Hybridní algoritmus založený na R a SciPy balících
Economic parameters: Software, Freeware
R&D Projects: GA ČR(CZ) GA16-24309S
EU Projects: European Commission(CZ) ERA-Net for Research Programmes on Rare Diseases No. 8F19001
Institutional support: RVO:61388971
Keywords : Biomolecular motifs discovery * counting and iteration
OECD category: Biology (theoretical, mathematical, thermal, cryobiology, biological rhythm), Evolutionary biology
https://peterslab.org/downloads.php

MatrixMotif is novel hybrid algorithm for biomolecular language processing allowing mining of hidden TFBS motifs from enriched NGS datasets. We primarily stacked on Hypergeometric Distribution Model allowing pseudo-random estimation of DNA motif seeds which are subsequently marginalized through eigenvalue of the Markov chain transition matrix. To get rid of local optima in our constraints we were iterating our motifs using property of Maximum Entropy to maximize our expectation. The maximum free motifs are then re-estimated using hidden Markov matrix and the SuperBinder motif is selected according its maximum log-likelihood score.
Permanent Link: http://hdl.handle.net/11104/0304632