Kriti Puniyani

• Learning Gene Networks from 2-D Images: We are trying to learn gene regulatory networks from 2-dimensional images of Drosophila embryos. Using a variety of techniques from computer vision, and machine learning, we are developing models to be able to determine expression patterns of genes and find the relationships between them.
• Transcriptional Control in Drosophila: We are trying to develop models that predict transcriptional gene expression, given the CRM sequence and the amount of transcription factors present in the cell. We have modeled the process of transcription control using thermodynamic principles, and have expressed it as a graphical model. Using approximate inference, we are able to fit and predict gene expression.
• Structured Association Mapping: We are trying to solve the classical problem of population structure in association mapping, using multi-variate regression. Via this project, I am trying to use various optimization techniques to optimize convex problems efficiently.