Comprehensive molecular profiling of Adenovirus receptor (CXADR) by the nference platform reveals strong expression in liver tissues: Implications for rising pediatric hepatitis cases in Europe and North America

Adenovirus infection and pediatric hepatitis: Pan-tissue molecular profiling of Adenovirus receptor (CXADR) protein reveals strong liver expression

Figure 1 - nferX Signals app query for "Adenovirus AND Receptor" across the core corpus (PubMed, Preprints, Clinical Trials, SEC filings, Patents etc) shows CXADR gene as strongest candidate for viral receptor from published biomedical literature.

Figure 2A - nferX Proteomics app shows that Liver is one of the most enriched human tissues according to available Mass Spectrometry proteomics data for CXADR protein expression.

Figure 2B - nferX Proteomics app shows that CXADR protein is also expressed at lower levels of enrichment in the respiratory tract (e.g. saliva, lungs) as well as the gastrointestinal system (e.g. colon, rectum)

Figure 2C - nferX Proteomics app details show 9 different mass spectrometry studies as all displaying normal and diseased liver tissues with enriched CXADR protein levels.

Figure 2D - nferX Proteomics app helps dive into 1 of the 9 individual studies with Mass Spec data, showing CXADR significantly expressed in 223 of 248 liver samples from both non-tumor (healthy) liver as well as hepatocellular carcinoma liver samples.

Figure 2E - nferX Proteomics app shows fetal brain and heart tissues  express the CXADR protein from mass spectrometry data suggesting pediatric population may also express CXADR protein in these tissues.

Figure 2F - nferX Proteomics app shows fetal liver, heart and ovary tissues express the CXADR protein from mass spectrometry data suggesting pediatric population may also express CXADR protein in these tissues.

Figure 3. nferX proteomics app has an immunohistochemistry (IHC) panel that shows that liver tissue stains moderately with a CXADR antibody (Source: Imaging data from Human Protein Atlas - HPA).

Figure 4a - nferX Single Cell app shows that Hepatic stem cells and Cholangiocytes have one of the most significant expression levels for the CXADR gene.

Figure 4b - nferX Single Cell app enables comprehensive profiling of over 3 million single cell RNA-seq (scRNAseq) human samples and provides distribution of 310K+ cell samples with non-zero expression of CXADR gene. This result shows that both the respiratory tract (nasal cavity, lungs) as well as gastrointestinal system (large intestine, small intestine) have cell types expressing the CXADR gene.

Figure 5a - nferX RNA Explorer app enables comparison of cohorts across studies for CXADR gene expression. Here Group A has White ethnicity samples (n = 6497) from 118 studies, being compared with Group B with 7 studies of African American/Black and Hispanic/Latino individual's samples (n=124).

Figure 5b - nferX RNA Explorer app shows individuals in Group B (African American/Black or Latino/Hispanic ethnicity) having modestly more expression of CXADR gene than Group A (White ethnicity) with Cohen's D value of 1.06 (effect size).

An image representation that’s a close analog for words in generative tasks

Reveals clues to its state-of-art performance in NLP tasks

COVID-19 questions — a use case for improving sentence fragment search

Progress in(slow) conscious task solving?

Also: some recent clues/insights into what makes BERT so effective

For sentence similarity/document search applications

Unsupervised creation of interpretable sentence representations

For select supervised tasks with self-supervised learning(SSL) models satisfying certain properties

T5 — a model that explores the limits of transfer learning

An approach to evaluate a pre-trained BERT model to increase performance

To create word representations capturing deep bidirectional context

XLNet — a clever language modeling solution

BERT is a prize addition to the practitioner’s toolbox

Swiss army knife for unsupervised task solving

A hybrid approach combining symbolic processing with distributed representations

A prerequisite to use a pre-trained model as is, without fine tuning

Quantitative evaluation of a pre-trained BERT model