Location: Yawkey Gallery
During the networking reception at the Women in Data Science Conference, students are invited to display research posters.
If you’d like to submit and display a poster, please fill out the Registration Form. Submissions are open to all women in data science, and students from undergraduate, graduate, or PhD programs.
Abstract Submissions for the Poster Session:
Image-based screening is a useful tool to understand complex cellular processes and to identify genes, proteins or small molecule compounds that regulate such pathways. Screening assays that monitor the function of a protein or pathway can visualize a change, such as a change in the fluorescent intensity of a reporter protein, and produce a large volume of image-based readouts. An example of this can be the increase in intensity of a fluorescently tagged protein due to differential biochemical and biomechanical stimuli. One transcription factor that is known to undergo such changes is the Kruppel-like factor 2 (KLF2) that has been characterized as a critical integrator of the vascular endothelium and its associated dysfunction(s). In the Guillermo García-Cardeña Lab at Brigham and Women’s hospital, we have used a KLF2-GFP lentiviral vector in human umbilical vein endothelial cells (HUVECs) to characterize the transcription factor and its response to blood flow characteristic of regions in the vasculature that are resistant to atherosclerosis (pulsatile, laminar flow) as well as its response to known pharmaceutical inducers of KLF2 (Simvastatin) through accurate, automated cellular (object) intensity measurement and analysis using CellProfiler 2.0 and R with statistically significant results using the two-way ANOVA, followed by the Holm-Å ÃdÃ¡k method. These results were validated using fluorescence-activated cell sorting (FACS). Ultimately, these assays and the corresponding modules developed in CellProfiler can be adapted to a variety of validation experiments and can be used in early stage drug discovery to identify compounds that modulate or inhibit KLF2 signaling and internalization.
The expansion and specialization of growing scientific communities has increased the challenge of finding appropriate peer reviewers for new scientific works, leading to the widespread adoption of automated paper-reviewer matching systems based on models of scientific expertise. Keywords extracted from scientific works of reviewers can behave as a summarization of their expertise and therefore be used to match papers to reviewers in order to find potential matches between the two. In this poster, I suggest several experimental model for successfully extracting relevant keywords from research articles and use these to match papers to their reviewers. I use bids provided by reviewers for submissions in the uai2017 conference as ground truth and evaluate the results using a thresholded recall score.
Empowering targeted tenant organizing: geographic forecasting of housing insecurity:
We map housing insecurity to empower targeted tenant organizing by the Philadelphia Tenants Union (PTU). The PTU does citywide canvassing to find tenants struggling with abusive landlords, help them to form unions and work collectively to secure safe and affordable housing. Using data made available by the city and the census, as well as real-time indicators of neighborhood change from social media, we construct models to direct the PTU’s long- and short-term activities. A target system will forecast evictions at high levels of geographic resolution for an interactive community empowerment tool focused on housing security.
This project aims to help the Red Cross predict flooding occurrences in Togo due to overflow in the Nangbeto Dam. Flooding is a result of both high flow rate and the water level in the dam at any point in time. This project focuses specifically on predicting the flow rate in the dam using precipitation data from eight locations around the country. A Lasso model and cross validation were employed to evaluate the significance of the predictors and capture the variance of flow rate.
Additional Authors: Jessica Wert, Oumayma Koulouh