The Gene Expression Omnibus (GEO) is a major open biomedical research repository for transcriptomics and other omics datasets. It currently contains millions of gene expression samples from tens of thousands of studies collected by many biomedical research laboratories from around the world. While users of the GEO repository can search the metadata describing studies and samples for locating relevant studies, there is currently no method or resource that facilitates global search of GEO at the data level. To address this shortcoming, we developed Rummageo, a webserver application that enables gene expression signature search against all human and mouse RNA-seq studies deposited into GEO. To enable such a search engine, we performed offline automatic identification of conditions from uniformly aligned GEO studies available from ARCHS4, and then computed differential expression signatures to extract gene sets from these signatures. In total, Rummageo currently contains 135,264 human and 158,062 mouse gene sets from 23,395 GEO studies. Overall, Rummagene provides an unprecedented resource for the biomedical research community enabling hypotheses generation for many future studies.

TargetRanger is a web-server application that identifies targets from user-inputted RNA-seq samples collected from the cells we wish to target. By comparing the inputted samples with processed RNA-seq and proteomics data from several atlases, TargetRanger identifies genes that are highly expressed in the target cells while lowly expressed across normal human cell types, tissues, and cell lines.

There is a rapid growth in the production of omics datasets collected by the diabetes research community. However, such published data are underutilized for knowledge discovery. To make bioinformatics tools and published omics datasets from the diabetes field more accessible to biomedical researchers, we developed the Diabetes Data and Hypothesis Hub (D2H2).

We use a deep learning model to detect the user's emotional state in real-time. Based on the user's emotional state, we apply a video filter to the user's face. We use Tensorflow to detect the user's emotional state and the OpenCV library and computer vision to apply the video filter.

A deep learning framework for 2d plant cell segmenation. This model was delevoped based off the work of Wolny et. al’s paper “Accurate and versatile 3D segmentation of plant tissues at cellular resolution.” The model archetecture exists in segmentor.py and the training and testing as well as visualaztion functions exist in assignment.py. Data is avaliable at https://osf.io/uzq3w and the specfic sets that were used in preprocessing and testing were the LateralRootPrimordia images in the test and train folders. With only a small set of these images a 97% accuracy was achieved.

This Appyter is inspired by the work of Bosse, Kristopher R et al. which compared neurobastomas vs normal tissue in GTEx to identify a promising candidate immunotherapeutic target. The goal is to allow rapid screening of targets with the help of normal tissue data from GTEx and GEO data through ARCHS4, as well as single-cell data from Tabula Sapiens and the Human Cell Atlas. The Appyter takes tumor expression data and attempts to rank significantly differentially expressed genes when compared with with either bulk RNA-seq data from GTEx or ARCHS4, or single-cell RNA-seq data from Tabula Sapiens or Human Cell Atlas, across all tissues. The Genotype-Tissue Expression (GTEx) Project was supported by the Common Fund of the Office of the Director of the National Institutes of Health, and by NCI, NHGRI, NHLBI, NIDA, NIMH, and NINDS. GTEx Version 8 gene counts was processed to produce gene summary statistics. ARCHS4 provides access to gene counts from HiSeq 2000, HiSeq 2500 and NextSeq 500 platforms for human and mouse experiments from GEO and SRA. We processed ARCHS4 Version 11 to produce gene summary statistics. The Tabula Sapiens dataset was created by the The Tabula Sapiens Consortium. We processed the Tabula Sapiens dataset to produce gene summary statistics. The Human Cell Atlas provides access to single-cell data contributed by the scientific community. We combined and processed 15 datasets from the Human Cell Atlas to produce gene summary statistics. Immunotherapeutic candidates must have limited expression in normal tissues to be considered safe targets, so proteomic visualizations of the highly expressed genes in normal tissues may be useful in assessing gene candidacy. Proteomics data were obtained from the Human Protein Atlas with IHC-based expression profiling, the Human Proteome Map with MS-based expression quantification, and a GTEx proteome project using TMT MS.