Ashish V Tendulkar
Ph. D.Research Scholar Wing,
KReSIT, IIT Bombay,
Powai, Mumbai-400 076
Email: ashish[AT]it.iitb.ac.in
More recent page Ashish Tendulkar @TIFR
| Home | Research | Curriculum Vitae |
Ashish V TendulkarPh. D.Research Scholar Wing, KReSIT, IIT Bombay, Powai, Mumbai-400 076 Email: ashish[AT]it.iitb.ac.in More recent page Ashish Tendulkar @TIFR |
|
| Research Topic | ||||||||||||||||||||||||
|
Analysis of Protein Structure using Geometric and Machine learning techniques
|
||||||||||||||||||||||||
| Research Summary | ||||||||||||||||||||||||
|
Understanding protein sequence, structure and function relationship
has been of considerable interest to the community of biological
scientists. The sequence-structure correlations enables computational
protein structure prediction from its sequence, thus bridging
the sequence-structure barrier. On the other hand, structure-function
relationship enables elucidation of protein function, which is the
most important task in the post-genomics era. In my Ph D thesis, I have
analyzed protein structures using geometric and machine learning techniques
with specific goals of protein structure prediction and functional
classification of the protein structures. More specifically, we represented
geometries of substructures in proteins with unilateral structure descriptors
in form of Geometric Invariants and applied clustering to form groups of
similar geometries. This approach enables efficient and scalable
all-against-all comparison between substructures and application of
standard data mining techniques. It, thus, overcomes the limitations
caused by inherent nature of pairwise comparison step employed in the
state of art techniques. We constructed a library of functionally
important constellations of small number of amino acids in protein and
a library of structurally similar local conformations in proteins.
Further, we constructed a visual map of protein local conformational space
and modelled it as a mixture of Gaussians. The component in the mixture
corresponds to a class of local conformations. We found total of 46
classes of local conformations with varying mixing proportion. The
sequence-structure relationship learnt for these classes is
exploited to predict protein structure using state of art sequence mining
techniques.
For detailed information about my research activities, publications, programs and results, visit Research Activities page. |
||||||||||||||||||||||||
| List of Journal/Conference Publications | ||||||||||||||||||||||||
|
||||||||||||||||||||||||
| Awards | ||||||||||||||||||||||||
|
||||||||||||||||||||||||
| Career Outline | ||||||||||||||||||||||||
|