Only keep genes above this level in at least n samples:

Min. level

n samples

Log Transformation

No

Yes

Constant c for started log: log(x+c)

Keep genes with minimal counts per million (CPM) in at least n libraries:

Min. CPM

n libraries

Transform counts data for clustering & PCA.

VST: variance stabilizing transform

rlog: regularized log (slow)

EdgeR: log2(CPM+c)

Pseudo count c:

Missing values imputation:

Gene median Treat as zero Median within sample groups

Plot one or more genes

Search processed data

Do not convert gene IDs to Ensembl.

Processed data

Converted counts data

High-resolution figure

?

Aspect ratios of figures can be adjusted by changing the width of browser window.

Most variable genes to include:

Gene SD distribution Interactive heatmap
Correlation matrix Sample Tree

---

Customize hierarchical clustering (Default values work well):

Color

green-black-red

Distance

Correlation

Linkage

average

Cut-off Z score

Center genes (substract mean)

Normalize genes (divide by SD)

Center samples (substract mean)

Normalize samples(divide by SD)

Do not re-order or cluster samples

Heatmap data High-resolution figure

?

Most variable genes to include

Number of Clusters

Re-Run How many clusters? Gene SD distribution t-SNE map

Normalize by gene:

Mean center Standardization L1 Norm

Enriched TF binding motifs
K-means data High-resolution figure

---

Pathway database

Remove redudant genesets

Visualize enrichment Enrichment details

?

Enriched pathways for each cluster

Methods

Principal Component Analysis

Multidimensional Scaling

t-SNE

Pathway Analysis of PCA rotation

Coordinates High-resolution figure

?

Identifying Differential Expressed Genes (DEGs). See next tab for details.

Method:

DESeq2 limma-voom limma-trend

Using the limma package

FDR cutoff

Min fold change

Select factors & comparisons

Venn Diagram

Gene lists

FDR & fold-changes for all genes

Figure

?

Numbers of differentially expressed genes for all comparisons. "B-A" means B vs. A. Interaction terms start with "I:"

Examine the results of DEGs for each comparison

Volcano Plot MA Plot Scatter Plot TF binding motifs in promoters Gene list & data High-resolution figure

---

Enrichment analysis for DEGs:

Enrichment tree Enrichment network Enrichment details Enrichment using STRING API

Also try ShinyGO

?

Enriched pathways in DEGs for the selected comparison:

Top Genes for selected comparison:

Select method:

GAGE GSEA (preranked fgsea) PGSEA PGSEA w/ all samples ReactomePA

Geneset size: Min.

Max.

Pathway signifiance cutoff (FDR)

Number of top pathways to show

Use absolute values of fold changes for GSEA and GAGE

Remove genes with big FDR before pathway analysis:

Pathway tree Pathway network Pathway list w/ genes

High-resolution figure

High-resolution figure

* Warning! The many combinations can lead to false positives in pathway analyses.

?

This interactive map shows DEGs on the genome. Red and blue dots represent up- or down-regulated genes, respectively. Mouse over to see gene symbols. Click and drag to zoom in.

Filters: FDR

Fold change

To identify genomic regions significatly enriched with up- or down-regulated genes, we can use PREDA. Very slow (5 mins), but may be useful in studying cancer or other diseases that might involve chromosomal gain or loss.

Run PREDA (5 mins)

Biclustering can discover genes correlated on subset of samples. Only useful when sample size is large(>10). Uses methods implemented in the biclust R package.

Most variable genes to include

Method:

BCCC QUBIC runibic BCXmotifs BCPlaid BCSpectral BCBimax BCQuest

Enrichment database

Download all biclusters

?

Enriched gene sets in selected bicluster

Genes in this cluster

Identify co-expression networks and sub-modules using WGCNA. Only useful when sample size is large(>15).

Most variable genes to include (<3001)

Soft Threshold

Min. Module Size

Choose soft threshold Heatmap

---

Edge Threshold

Top genes

Change network layout

Enrichment database:

Download all modules Download network for selected module

The network file can be imported to VisANT or Cytoscape.

?

Enriched pathways among all genes in selected module

Email us for questions, suggestions, or data contributions. Stay connected via user group or Twitter. Visit our GitHub page to see source code, install a local version, or report bugs and request features. iDEP is being developed by a very small team: Dr. Xijin Ge and a graduate student (Homepage).

R as in Reproducibility

Documentation site. Source code on GitHub, where users can also report bugs or requet features.

To improve reproducibility, iDEP generates custom R code based on your data and choices of parameters. Users with some R coding experience should be able to re-run most analyses by downloading all of the files below. If Ensembl IDs is not used in users' original file, we should use the converted data file. Click through all the tabs and then download all these file to a folder. Run the Customized R code or the Markdown file. R Markdown example.

Customized R code Customized R code(Markdown) iDEP core functions Gene Info file Pathway file (large)

Previous versions of iDEP

iDEP 0.85 with Ensembl 95, archived on May 19, 2019
iDEP 0.82 with Ensembl BioMart version 92, archived on March 29, 2019
iDEP 0.73 with Ensembl BioMart version 91, archived on July 11, 2018

Citation

Please cite: Ge SX, Son EW, Yao R: iDEP: an integrated web application for differential expression and pathway analysis of RNA-Seq data. BMC Bioinformatics 2018, 19(1):534. PMID:30567491 Full text

Usage Statistics

As of June 5, 2019, iDEP website has been visited 30,895 times by 8,252 users from 91 countries. Fully functional only in early 2018, iDEP has been used by researchers to analyze transcriptomic data from sun flowers to primates to human, as demonstrated by the 22 papers citing iDEP.

Changes

v 0.38 Gene ID conversion, remove redudancy; rlog option set to blind=TRUE.

v 0.39 Reorganized code. Updated to Bioconductor 3.5; solved problems with PREDA 9/8/17.

v 0.40 Moved libraries from the beginning to different places to save loading time.

v 0.50 Enable changing colors for heatmap. Fixed error with voom; also changed method to TMM, see here.

2/5/2018 V 0.62 : Add tree and networks to visualize overlaps between enriched gene sets, in K-means, DEG2, and pathway tags. Downloads of pathway analysis results and high-resolution figures.

2/6/2018: Fixed errors caused by gene symbol matching for unknown species. More user control of hierarchical clustering tree

2/9/2018: V 0.65 Added API access to STRINGdb website on the DEG2 tab. Supports thousands of bacterial species

2/10/2018: V 0.66 Improved API access to STRINGdb, by adding automatic species matching.

2/11/2018: V 0.67 Tested with larger dataset of 259 samples. Changed figure configurations.

2/14/2018: V 0.68 Fixed Pathview loading code. Connected Pathview to PGSEA.

2/25/2018: V 0.68 Fixed Fold-change, FDR data upload and parsing. Figure resolution using the res=150 option help improve readability of labels.

2/28/2018: V0.69 Change interactive heat maps to re-order columns

3/12/2018: V0.70 Generating R code and downloading annotation files used in analysis.

3/14/2018: V0.71 Improve R markdown file; add color to EDA plots; detect bias in sequencing depth

3/18/2018: v0.711 Fixed error caused by gene names containing characters such as ' or "

3/28/2018: v0.712 Fine tuned EDA plots

4/3/2018: V0.713 add permutations for fgsea. Expand quotes.

4/13/2018: V0.72 fixed bug caused error in R Markdown file when users choose species other than the default.

4/25/2018: V0.73 Fixed bug for GO terms tree in k-Means. Added layout button for network viz of GO terms

5/28/2018: V0.73 Enabled download of all FDR and fold-changes at the DEG1 tab. Problem with the FDR and fold-change data with only one columns.

7/28/2018: v0.80 New v0.80 Updated annotation database. Comprehensive pathway database for human. TF binding motifs for 200+ speceis. Old version made available.

7/29/2018: v0.80 Orgized UI.R according to Google R style guide.

7/30/2018: V0.80 Fixed error in downloaded up- and down-regulated gene lists. We thank Juan Xie for pointing this out.

12/2/2018: v0.81 High resolution figure download with eps format, which can be eidted with Adobe Illustrator.

3/5/2019: v0.82 Fix a bug regarding limma for identification of D.E.Gs. Up- and down-regulation are opposite in some cases.

3/29/2019: v0.85 Annotation database upgrade. Ensembl v 95. Ensembl plants v.42, and Ensembl Metazoa v.42.

5/19/2019: v0.90 Annotation database upgrade. Ensembl v 96. Ensembl plants v.43, and Ensembl Metazoa v.43. STRING-db v10

In loving memory of my parents. X.G.