Tutorial

What is IIIDB?

IIIDB is the first database that allows users to upload their own gene expression data for isoform-isoform interaction prediction. In IIIDB, the users can easily access high-confidence and low-confidence predictions of human IIIs, and then see the full evidence values for each predicted III (Figure 1). The IIIDB allows users to easily search IIIs and isoform modules, and then provides the evidence that led to each III prediction. Interestingly, the different isoforms within the same gene may be involved with different isoform modules that may open a new door to study differential functionality of isoforms of the gene. The IIIDB also provides the comprehensive enrichment analysis results (GO and pathway) for isoform modules, which may help biologists to study the biological insight of network modules in isoform level.

Figure 1. The overview of IIIDB.

Isoform coexpression network construction

To construct the Isoform coexpression networks, we selected 14 RNA-seq datasets with at least 10 experiments from Sequence Read Archive (SRA) database (Table 1). We performed the eXpress with Bowtie2 aligner to obtain isoform expression values. NCBI Reference Sequences (RefSeq) mRNAs with protein sequences was used as transcriptome annotation, which included 31,454 RefSeq IDs (Jan 2013 version).
Table 1. RNA-seq datasets.

Logistic regression model

We have assembled a collection of proteomic and genomic data sources for predicting III (See Figure 2) as follows: (1) RNA-seq data: We collected 14 RNA-seq datasets, each with sample size at least 10. Pearson’s expression correlations for all isoform pairs were derived from each dataset. We hypothesize that two frequently co-expressed isoforms are more likely to interact than those that are not. (2) Domain-domain interaction data: we set domain score is how many domain pairs we find from iPfam database.


Figure 2. We performed logistic regression with 14 RNA-seq datasets and the domain-domain interaction database to construct IIIDB.

Package

IIIDB provided a package using R language. Users can use this package to process their own gene-expression data to perform isoform-isoform interaction prediction.

Download Package

Prepare:
Unzip package file and put files in the package into the same folder. Then, you will get the following files:


Input file format:

test_file.csv

ID1_ID2 - isoform-isoform pairs as RefSeq id and separated by ":" (e.g. NM_000014:NM_000373).
exp_d1_score - the absolute values of Pearson correlations for the isoform pairs derived from first RNA-seq dataset.
exp_d2_score, exp_d3_score, ... - the absolute values of Pearson correlations for this isoform pairs derived from other RNA-seq datasets. In this example we use five RNA-seq datasets.

Use steps:
#setup your path, for example: unzip package file to D:/IIIDB
setwd("D:/IIIDB")
# execute "code.R", please wait for a moment.
source("code.R")


Output file:
All prediction result will output as a file named output_result.txt.

output_result.txt

IIIDB functions:

There are three main functions in IIIDB: (1) Upload your gene expression data (2) Interaction search and (3) Module search. The searching function has three parts: high-confidence interaction prediction search, low-confidence interaction prediction search, and isoform module search. The IIIDB provides auto-complete function in all search functions. Users can input a gene symbol or gene ID into the auto-complete field that provides quickly searching and selecting the partially matched terms. Furthermore, by entering more characters, it will filter down the list to better matches. The auto-complete function not only helps the users search efficiently but also makes a quick filtering.


Figure 3. The screenshots of main functions in the IIIDB.

1) Upload Expression Data

Users can upload their own gene expression data for isoform-isoform interaction prediction. IIIDB present top 100 isoform-isoform interactions and use these interactions to construct a Top 100 III network. Users can get complete prediction result as a file.


1.1) Upload your gene expression data

Upload your gene expression data to do isoform-idoform interaction prediction.

Figure 4. Upload your gene expression data.
Upload progress of your gene expression data.

Figure 5. Upload progress.
The screenshots of server processing.

Figure 6. The scrddnshots of server processing.

1.2) Data format

Recommended data formats for expression data are separated by delimiters, if IIIDB can not process you data, please check your data fomat again.


Figure 7. Data format of upload gene-expression data.

1.3) Interaction prediction result

Interaction prediction result of your data.

Figure 8. Interaction prediction result table.

1.4) Download prediction result as a file

IIIDB just present top 100 isoform-isofrom interactions as tabular data. Then, you can get complete prediction result as a file.

Figure 9. Download all prediction result as a file.

1.5) Top 100 III network

IIIDB using top 100 isoform-isofrom interactions to construct a "Top 100 III network".

Figure 10. Top 100 III network.

2) Interaction Search

In interaction search function, users can input a gene symbol or gene ID to search associated IIIs and isoform modules. The interaction search section provides interface on searching high-confidence (score > 2.575) or low-confidence (score > 1.692) prediction. The resulting III table for interaction search shows full evidence values including Pearson correlations for 14 RNA-seq datasets and domain interaction score (Figure 14). Users can input a gene symbol or gene ID into the auto-complete field that provides quickly searching and selecting the partially matched terms. Furthermore, by entering more characters, it will filter down the list to better matches. The auto-complete function not only helps the users search efficiently but also makes a quick filtering.


Figure 11. Users can input gene symbol or gene ID into the auto-complete field.

2.1) High-confidence interaction prediction search


Figure 12. High-confidence interaction prediction search.

2.2) Low-confidence interaction prediction search


Figure 13. Low-confidence interaction prediction search.

2.3) Interaction search result


Figure 14. Interaction search by high-confidence prediction.

3) Module Search

The resulting page for module search includes the user friendly network graph and pathway/GO enrichment analysis result (Figure 16, 17, 18).

Figure 15. Module search input field.

3.1) Module network


Figure 16. Module search result.

3.2) Pathway enrichment analysis


Figure 17. Module search pathway enrichment analysis result.

3.3) GO enrichment analysis


Figure 18. Module search GO enrichment analysis result.

Cytoscape web

We developed an interactive network web interface to visualize the isofrom-isofrom interactions using CytoscapeWeb (http://cytoscapeweb.cytoscape.org/) (PMID: 20656902). 

Layout

Networks provide five layouts:
  • ForceDirected
  • Circle
  • Radial
  • Tree
  • Compound

Export

Networks can be exported in six three formats:
  • Scalable Vector Graphic (SVG)
  • Portable Network Graphic (PNG)
  • Simple Network File (SIF)