Leveraging spatial, anatomical, microarray expression data from the Allen Institute for Brain Science. This program maps genes of interest (Goi) by their levels of expression to brain regions (Roi). The identified regions that co-express the input Goi can, in turn, be studied further with many experimental modalities. This program can be customized to map gene expression data and identify significant Roi from similar reference expression data in other contexts outside of neuroscience.
Technologies in experimental neuroscience have exponentially improved, however, translational neuroscience (from mouse-to-humans) has not been as successful in producing therapeutics that treat human disease. Although there are many limitations to research performed with human participants, one of the main advantages is an easier translation of findings into clinical settings. However, current tools lack the ability to query molecular and brain-region specific data, an approach that is required for bedside-to-bench-to-bedside research.
Goi2Roimapping utilizes a compilation of ex vivo, spatial, microarray expression data from the human brain that is freely available at the Allen Institute for Brain Science. Goi2Roimapping allows a user to input Goi relevant to their field of study (eg. targets of microRNAs, genes detected through GWAS, etc.). Once Goi are defined by the user, Goi2Roimapping will iterate though brain regions and compare expression levels of the Goi relative to all other genes (using a Wilcoxon test). At the end, the user will obtain brain Roi where those genes are significantly co-expressed as a .csv file. Additionally, box plots showing relative expression of the Goi relative to all other genes are generated. By determining brain regions where input genes are expressed, investigators can take on exploring different aspects of regional brain function in various modalities, including human brain imaging. By determining whether Goi are expressed in specific brain regions, investigators can better inform their current and future experiments. Ultimately, my aim is that this tool can help neuroscientists bridge the gaps in molecular neuroscience.
With minimal experience with other programming languages, I consider this project my first introduction into programming in general, and certainly my first introduction to Python. Therefore, I've leveraged several important basic libraries:
pandas – to iterate through dataframes and create new output dataframes.
matplot.lib - to generate boxplots
scipy.stats. - to perform comparisons between input goi and all other genes as we iterate through the dataframe analysis.
This program requires Python 3.8 or higher.
In its current form, the program asks the user to input a working directory for the reference data and brain region key. The program will ask the user for this at key steps when required and then ask for further required inputs (gene names of interest). Note that currently, the program is designed to iterate through a dataset containing median expression values from microarray data of 6 compiled reference brains. Next, (and soon) I plan to update the program to add an additional output: mapping the identified brain regions onto a glass brain using Nilearn.