Amrita Roy Choudhury was awarded for the best poster at CMTPI (‘Computational Methods in Toxicology and Pharmacology integrating Internet Resources’), in Istanbul 4. -8. July 2009. The encoding of protein sequences and building of model for classification of transmembrane segments of proteins.
E.02 International awards
COBISS.SI-ID: 4193562The information about the structure of trans-membrane segments of proteins of known 3D structure is exploited to predict the trans-membrane domains of structurally unresolved target protein. With the help of known structures the trans-membrane domains are encoded and a model for prediction of trans-membrane segments is built. Four transmembrane alpha helices were predicted in bilitranslocase. This result is partially confirmed with experimental studies using particular antibodies corresponding to parts of amino acid sequences of bilitranslocase.
B.04 Guest lecture
COBISS.SI-ID: 4558618PhD thesis included the study of inhibition of bilitranslocase, a trans membrane protein, whose primary function is the transfer of bilirubin from blood to liver cells. We have tested the interaction with purine, pyrimidines and their derivatives. Information on chemical structure and inhibition constants were used to construct models with artificial neural networks. Correct classification of compounds and good predictive ability was achieved by counter-propagation neural network model coupled with a genetic algorithm; RMS error for the validation set was 0.47 log units of binding affinity.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 250510848We present an approach towards structure elucidation of bilitranslocase (BTL) transmembrane regions. BTL is a membrane protein which transports bilirubin from blood to liver cells. The sequence and secondary structure information of transmembrane segments of proteins with known 3D structure is exploited to predict the transmembrane domains of structurally unresolved target protein. With the help of known structures the trans-membrane domains are encoded in such a way that it is possible to group and classify them with respect to their specific sub-structural characteristics and to build a model for prediction of transmembrane segments. In order to explore the bilitranslocase transport mechanism, we tested a set of non-congeneric compounds for their competitive inhibition constants in the investigated protein-substrate system. The information about chemical structure of small molecules that inhibit bilitranslocase helps us to build a hypothesis about the transport mechanism of the studied biological system.
B.04 Guest lecture
COBISS.SI-ID: 4708634Despite the importance of transmembrane proteins and growing interest in them, the vast majority of the membrane proteins remains underexplored owing to experimental difficulties. To fill this knowledge gap, several in-silico methods are developed aiming to predict the transmembrane regions, topology and structure. Although prediction for [alpha]-helical transmembrane regions can be made with considerable accuracy, it is not so in case of transmembrane [beta]-strands. The shorter and less hydrophobic transmembrane [beta]-strands are much harder to predict. The [beta]-barrel transmembrane proteins are present in the outer membrane of bacteria, cell organelles like mitochondria and chloroplasts. They function as ion transporters and play role in passive nutrient uptake. In this work, we present a data-driven prediction model of J3-strand transmembrane region. The prediction is done based on amino acid sequence information without using any evolutionary data from multiple sequence alignments. Data on [beta]-barrel transmembrane proteins with atomic resolution structures and known transmembrane region is collected from public domain databases PDB and PDBTM. The protein sequences are separated into their transmembrane and non-transmembrane regions. The model is developed based on non-linear counter-propagation artificial neural network using mathematical descriptors defining the transmembrane protein sequences. The model shows 83% prediction accuracy when tested with external validation set. To further improve the prediction for unknown protein sequences and successfully eliminate false positives and negatives, statistical data on amino acid distribution in transmembrane [beta]-strands is incorporated in the final prediction. Finally, we did a benchmarking study comparing our developed prediction method with other algorithmic techniques and predictors available.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 4777242