Identification of disease markers in curated projects

Note: This is an ongoing effort and the results are available for a limited number of projects; more will be added as analyses proceed.

Overview

Marker identification was performed for selected projects referred to as curated projects in GMrepo.
These projects were selected based on the following criteria (in addition to available manpower and time):

• High-quality meta-data with clearly defined diseases and healthy controls
• High-quality sequence data
• Sufficient numbers of both cases and controls, usually more than 10 per group

Marker taxa

A marker taxon refers to a genus or species that shows significantly different abundances between two groups
(e.g., cases vs. controls or disease stages such as adenoma vs. CRC), identified using
LEfSe (Linear discriminant analysis effect size)
PMID: 21702898

Marker taxa were identified on a per-project basis.
Please consult the All marker taxa page for more information.

Curated projects

In GMrepo, manual curation was performed for selected projects in order to:

• Select usable runs with clearly defined phenotypes
• Merge multiple runs corresponding to the same sample and calculate taxon abundances per sample (instead of per run)
• Group samples according to their phenotypes
• Identify marker taxa between phenotype groups of interest, e.g., CRC vs. Health

In addition to project meta-data and run tables, curated project pages also include:

• A list of curated samples and their corresponding phenotypes and runs
• Marker taxa identified between pairs of phenotypes (referred to as phenotype comparisons)

Marker taxa are shown in both table and boxplot views.

For curated projects with more than two phenotype groups, e.g.:

• Healthy controls
• Adenoma
• Colorectal cancer (CRC)

Multiple phenotype comparisons can be performed, such as:

• Healthy control vs CRC
• Healthy control vs Adenoma
• Adenoma vs CRC

This enables identification of marker taxa across different disease stages.