Skip to content

Commit 834ce0f

Browse files
authored
Update README.md
1 parent e7f30a3 commit 834ce0f

File tree

1 file changed

+1
-3
lines changed

1 file changed

+1
-3
lines changed

README.md

Lines changed: 1 addition & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -2,9 +2,7 @@
22

33
![Workflow](./process.svg)
44

5-
## Project: `A deep learning method for quantitative immune defense function evaluation and its clinical prognostication`
6-
7-
The immune defense function protecting body from invasive patho-gens is a key indicator of health. However, it lacks of methods for quantita-tive and precise evaluation. This study introduces ImmuDef, a novel algo-rithm that provides a precise and quantitative assessment of anti-infection immune defense function based on RNA-seq data. ImmuDef employs the following pipeline: (1) transforming gene expression into robust immune signatures of MSigDB C7 immunologic gene sets; (2) selecting immune re-lated features through comparisons of AIDS vs. healthy controls (HC) and dead-survived sepsis patients; (3) dimensionality reduction and latent space construction via a variational autoencoder model, QImmuDef-VAE. Build-ing on QImmuDef-VAE’s latent representation, a defense immune score (DImmuScore) was calculated by measuring the distance between a patient and HC within the latent space, using healthy individuals as reference. We validated ImmuDef on 3,202 samples across HC and different infectious diseases, which were stratified into four immune states: immunodeficiency, immunocompromised, immunocompetent, and immunoactive. As a result, DImmuScore calculated by ImmuDef achieves high classification accuracy (AUC: 0.79-1.00) among samples with various immune states and infection. And in AIDS patients, lower DImmuScore predicted mortality, aligning with immune reconstitution dynamics. Furthermore, DImmuScore can serve as a metric for disease severity across hepatitis B, tuberculosis, and COVID-19 cohorts, where its gradient directly quantifies disease progression. As an ap-plication, the DImmuScore can be a strong prognostic indicator, effectively stratifying mortality/survival in both sepsis and COVID-19 patients with same symptom. These results indicate that ImmuDef can accurately reflect immune function and disease states of infectious diseases. By applying deep learning on immune pathway features, we designed the first algorithm for quantitatively calculating immunity of anti-infection. It enables accurate evaluation of the immune defense function, offering clinical potential for disease surveillance and prognosis prediction.
5+
## Project: `A deep representation learning method for quantitative immune defense function evaluation and its clinical applications`
86

97
# Requirements
108
## Python Package Requirements

0 commit comments

Comments
 (0)