picture.jpg
Aaditya V. Rangan


Courant Institute of Mathematical Sciences, New York University,
251 Mercer Street, New York, NY 10012.
Phone: (212) 998-3303, email: rangan (at) cims (dot) nyu (dot) edu
webpage: http://www.cims.nyu.edu/~rangan/
Office Hours Fall 2015: Thursday 1:30pm-3:25pm
Room: 1123 WWH

Research Interests: Applications of numerical-analysis and scientific-computing to the biological sciences.

CV (last updated 2024/04/30)

List of publications (last updated 2024/04/27)


Cryo-Electron Microscopy (Cryo-EM)

Recently I've been exploring computational strategies for use in cryo-em.
Link to paper [robust ab-initio reconstruction with small data-sets]
Link to paper [radial compression for alignment]
Link to paper [factorization of the translation operator]
These tools (and more) are currently available at (Link to github ).
A more user-friendly version (in Python) is in development, and will be linked as soon as possible.


Bioinformatics

I've also been applying a 'loop-counting' algorithm for biclustering genomic data. I believe the results are rather promising.
Link to paper [analysis of genotyped data: Bipolar Disorder]
Link to paper [analysis of gene-expression data: Alzheimer's Disease] with additional benchmarking in the Appendix.
Link to paper [analysis of genotyped data: Alzheimer's Disease]

Here is a brief explanation of what the loop-counting algorithm actually does, as well as a tutorial which presents an example of this algorithm applied to a standard data set:
Link to paper
(tutorial on github)
A much more detailed explanation of this loop-counting algorithm is given in:
(tutorial in pdf)
(Supplementary Material).
A comparison of loop-counting with other popular clustering algorithms (i.e., U-MAP and Louvain clustering) is discussed here.
This comparison (taken from the Appendix linked above) demonstrates the advantages of loop-counting as applied to single-cell RNA-sequencing data.
A significantly more efficient implementation of this algorithm (written in C) is available at:
lakcluster_ver18 on github
This implementation also includes several subroutines which perform binary vector-vector, matrix-vector and matrix-matrix operations.

I've also worked on simple strategies for fitting time-series data with stochastic-differential-equations (SDEs), allowing for certain kinds of causal inference (Link to github).
This method is described in more detail here: Link to paper