The SMART (Shiny Methylation Analysis Resource Tool) App is a user-friendly and easy-to-use web application for comprehensively analyzing the DNA methylation data of TCGA project.
The data used in the SMART App are directly pulled down from UCSC Xena public data hubs (https://xenabrowser.net) upon users’ request. Some minor processing is performed to ensure the consistency of the sample names.
DNA methylation (Methylation450K) (n=9,639) Pan-Cancer Atlas Hub
TOIL RSEM tpm (n=10,535) UCSC Toil RNAseq Recompute (Gene expression RNAseq)
TOIL RSEM tpm (n=10,535) UCSC Toil RNAseq Recompute (Transcript expression RNAseq)
Curated clinical data (n=12,591) Pan-Cancer Atlas Hub
Gene-level copy number (GISTIC2 thresholded) (n=10,845) TCGA hub
Gene level non-silent mutation (n=9,104) Pan-Cancer Atlas Hub
This is the SMART App start screen:
The home page displays the number of DNA methylation samples available from TCGA project
and provides a quick search interface.
Users can enter a gene symbol (e.g. TRIM58) into the ‘Quick Start’ box to search for a gene of interest.
By clicking the “Go” button, a circular plot showing the chromosomal distribution of all associated CpGs of the input gene will be displayed.
To help users gain more useful information about the CpGs and their genomic locations along with transcripts, a detailed segment plot highlighting the transcripts, exons, UTR, CDS, CpG island regions, shelves and shores is displayed by clicking the
"Click to view the detailed chromosomal distribution". The name and the type of each transcript is given.
The genomic length is shown below. By default, the distance between any adjacent two lines stands for 1k. Users can set the distance scale.
The yellow arrow at the top stands for the strand direction, that is, towards right, +, towards left, -.
The coverage of the CpGs islands are displayed as the red regions.
Clicking the "Click to check probe-level methylation" will display a table showing the detailed information these probes, and users can select one of these probes to view its pan-cancer methylation profile and identify dys-methylated sites for further analysis.
Clicking the "Click to check gene-level methylation" will display the gene-level pan-cancer methylation profile.
The SMART app allows users to set custom cut-off values for a given cancer type to dynamically obtain differentially methylated CpGs and their chromosomal distribution.
When all parameters are set, the chromosomal distribution plot and the list of differentially methylated CpGs based on input parameters will be displayed.
The SMART App offers a function
that allows users to flexibly visualize CpGs. Users can select CpGs that you want to plot from the table,
check the corresponding chromosome locations,
select chromosomes that you checked from the above table in the panel, and
click the plot button, a beautiful circular plot showing the selected CpG sites and relative density will be displayed. If no chromosome is selected, the SMART App will automatically display all chromosomes.
This module provides functions for users to comprehensively
analyze DNA methylation taking other omics data and clinical stage into consideration.
Users can select multiple probes at the same time for easier visibility and interpretation.
The returned box plots will display all the selected probes plus an aggregation box plot showing the mean/median methylation of all the selected probes.
The first panel generates custom box plots for users for compare CpGs of genes between normal and tumor samples in a given cancer type.
The second panel plots methylation by pathological stages based on the TCGA clinical data. Two options are available, namely, major stage and sub-stage. For example, if users choose major stage for plotting, stage IIA/IIB will be included into the stage II group.
The third panel offers a function for plotting box plots comparing methylation between mutation and wild type groups.
The fourth panel provides a function for researchers to study the possible association between CNV and DNA methylation.
For correlation analysis, one can choose to analyze the correlation
at gene-level or transcript-level. When analyzing the correlation at transcript-level,
a segment plot highlighting the genomic locations of the transcript and CpGs will be displayed and the distances of each probe to TSS
will also be shown in the table below for users to locate the ones at the promoter region
Users can select probes from the table to visualize the correlation between expression and methylation.
The results are displayed as scatter and distribution plots. In distribution plot, each bar represents a sample, the names of the gene/transcript and CpGs are shown on the right,
the methylation and expression values are shown on the left. The samples are reorders according to the expression value.
For Survival analysis,
the SMART App offers univariate Cox regression analysis. Users can copy and paste a list of CpGs into the box,
select the cancer type of interest to conduct Cox regression analysis.
The hazard ratio, 95% confidence interval, z-score and p-value will be given.
Once users have identified the significant variables, they can use SMART app to draw survival curves.
The thresholds for high/low methylation level cohorts can be adjusted by users.
All data in the SMART App are accessible.
Figures are rendered as Portable Document Format (PDF),
which can be edited by Adobe Illustrator.
The development of the SMART App is ongoing and we intend to add more functions and promote results visualization.
Comments and suggestions from users for the future development of the application are greatly appreciated.