Posters - Schedules

There has been growing concern about police brutality in recent years in the US, especially as it disproportionately affects black people and other people of color. The Black Lives Matter move ment began in the wake of Michael Brown’s death in 2014, and reached a new level of national recognition in May 2020 after George Floyd was murdered. A central element of this is when civilians are fatally shot by police officers in the line of duty. In this report, we study the pattern of fatal police shootings in the United States by analyzing the Fatal Force database, compiled by the Washington Post [Tate et al., 2021]. Other police incidents, such as shootings while the victim was in custody or civilians killed by police officers by means other than shooting are not available in this data. By using this data and additional data about US population, we perform an analysis of this dataset. We determine the demographic and geographical factors associated with police shootings and conduct exploratory analysis into the change in shooting rates since the murder of George Floyd. Additionally, we fit a prediction model based on the historical data.

Science journalism shapes the public’s view of scientific findings and legitimizes sources as experts. Even if unintentional, biases may influence who is identified and ultimately included as an expert. To identify possible biases, we analyzed 22,001 non-research articles published by Nature. Our analysis considered two possible sources of disparity: gender and name origin (a proxy for author nationality). We extracted cited authors’ names as well as names of quoted speakers to predict gender and name origin.

To quantify the disparity in representation between science journalism and scientific publications, we chose first and last authors within primary research articles in Nature and a subset of Springer Nature articles in the same time period as our comparator. We found a skew towards male quotation in Nature science journalism-related articles, but is trending toward equal representation at a faster rate than academic publishing. Our name origin analysis found a significant over-representation of names with predicted Celtic/English origin and under-representation of names with a predicted East Asian origin in extracted quotes. Through our comprehensive analysis, we were able to quantify how recognized persons in news journalism vary by name origin and gender and compare these rates to scientific publishing background rates.

CRISPR/Cas9 is a genome editing tool widely used in biological and clinical research. Naturally-occurring human genomic variations, through altering the sequence context of CRISPR/Cas9 targets, can significantly affect its editing outcomes. However, these effects have not been systematically studied or documented. Herein, we present CROTONdb, a database that enables fast and comprehensive investigations of single nucleotide variant (SNV) effects on CRISPR/Cas9 outcomes. CROTONdb leverages state-of-the-art machine-learning predictors to evaluate unbiased SNV effects on CRISPR/Cas9 outcomes. As a case study, we investigated a candidate Cas9 target in oncogene FGFR3 that is sensitive to SNV effect of rs2305181. The minor allele of rs2305181 was predicted to substantially decrease the knockout efficacy from 83.8% to 62.5%. Importantly, rs2305181 has a minor allele frequency of 29.7% and 0.7% in the African and European populations, respectively. If CRISPR/Cas9 is programmed to target this region in primary human cells or in clinical settings, the population stratification of this Cas9-altering variant may lead to health disparities. Similarly, we predicted large SNV effects in Cas9 targets used in preclinical animal experiments and early-phase clinical trials. This highlights the ubiquitous Cas9-altering SNV effects that are made easily accessible by CROTONdb.

Twitter is one of the most popular microblogging and social networking services, where users can post, retweet, comment, and engage in collaborative discussions. However, improper usage of Twitter can be detrimental to science and even have a negative impact on mental health. Thus, analyzing tweets and Twitter data of various researchers will help us to deduce appropriate ways of using Twitter to advance in our research careers. Existing literature has analyzed the activity of scientists on Twitter, such as studying the relationship between Twitter mentions and article citations, determining the benefits of Twitter in the development and distribution of scientific knowledge, relevant metrics for prediction of highly cited articles, and type of content that researchers tweet etc. Most of the existing literature analyzed a limited number of researchers, compromising the generalizability of derived results. In our study, we have taken a comprehensive and systematic approach to analyze 167,000 scientists who published research papers on PubMed using data-driven methods. We observed various parameters like number of followers, number of friends, citation count and K-index, in the light of gender, ancestry and profession of the researchers.