Rheumatoid Arthritis: Public Transcriptome Data Analysis with Omics Playground

Published on May 20th, 2020
Written by Axel Martinelli
⏱ 3 min read

Rheumatoid Arthritis

Introduction

Today we will take a look at a study on Rheumatoid Arthritis published in 2017 on Arthritis Research & Therapy using our Omics Playground Platform.

In this study, the authors sequenced the transcriptome of 24 patients (12 healthy individuals, 5 patients with untreated RA and 7 patients undergoing treatment for Rheumatoid Arthritis) and proposed three novel candidate genes (namely ERBB2, TP53 and THOP1) as being involved in the pathogenesis of RA.

Clustering analysis

We started by performing a clustering analysis using the umap approach, but even when considering only the top 100 discriminative genes, we could not see any clear separate clusters, although most healthy samples did appear to congregate separately from most Rheumatoid Arthritis samples (Figure 1).

UMAP-based clustering of the samples using the top 100 most discriminative genes.
Figure 1. UMAP-based clustering of the samples using the top 100 most discriminative genes.

Differential expression analysis

Next we performed a Differential Expression analysis. Due to the limited amount of clustering, we used only one algorithm (DESeq2) with FDR<0.05 and Log Fold Change >0.5.

The results indicated that while there were extensive differences (1,085 DE genes) between the healthy patients and treated Rheumatoid Arthritis patients (probably due to the effect of the treatment), differences were more modest (only 195 DE genes) when comparing healthy and untreated patients (Figure 2).

Volcano plots of the DE genes in the pairwise comparisons of healthy vs treated RA samples (A) and healthy vs untreated RA samples (B).
Figure 2. Volcano plots of the DE genes in the pairwise comparisons of healthy vs treated RA samples (A) and healthy vs untreated RA samples (B).

Of the 11 genes that had been previously identified as being associated with Rheumatoid Arthritis and that the article indicated as unidirectionally DE in patients, only SLC22A4, PPIL4, IL2RB, EOMES, B3GNT2 and ANXA3 were significantly DE with the parameters we used when healthy samples were compared against treated Rheumatoid Arthritis sample and only PPIL4 when healthy samples where compared against untreated Rheumatoid Arthritis samples.

When it came to the three novel candidate genes for Rheumatoid Arthritis, no statistically significant differences in expression could be observed when the untreated patients were compared against the healthy controls, reflecting the similar expression levels (Figure 3). 

Differences in expression between healthy samples and treated samples were greater (Figure 3), but still not statistically significant for FDR<0.05. This could reflect the observation that differences between Rheumatoid Arthritis patients and healthy individuals, although statistically significant, were not particularly pronounced in some follow up experiments done by the authors in the same article.

Normalised (log2CPM) gene expression levels across the three sample groups for genes (A)TP53, (B)ERBB2, and ©THOP1
Figure 3. Normalised (log2CPM) gene expression levels across the three sample groups for genes (A)TP53, (B)ERBB2, and ©THOP1

Drug connectivity analysis

To round up the analysis, we compared the gene expression profile of the comparison between untreated and treated Rheumatoid Arthritis samples against the Drug Connectivity Map Database, in order to see if the profile generated by treatment was positively correlated with profiles from drugs with the anti-Rheumatoid Arthritis potential. 

The four most positively correlated modes of actions included histone deacetylase (HDAC), NfkB pathway, and heat shock protein (HSP) inhibitors, as well as retinoid receptor agonists (Figure 4). All of these targets have been associated with the development of rheumatoid arthritis in the literature, as shown in the links above.

Most positively and negatively correlated drug modes of action with the DE gene profiles of the untreated vs treated RA patient samples. The top four modes of action correspond to genes or pathways involved in the development of RA.
Figure 4. Most positively and negatively correlated drug modes of action with the DE gene profiles of the untreated vs treated RA patient samples. The top four modes of action correspond to genes or pathways involved in the development of RA.

Overall, we could not quite replicate the results obtained in the paper. We did observe differences in expression in the novel genes described by the article, although such differences were not statistically significant. Indeed, we could not find any mention in the literature (via a Pubmed search) of any of these genes in connection with Rheumatoid Arthritis after the publication of the article, suggesting that the findings have not been replicated since.

Scientist with lens exploring the drug connectivity analysis tab in Omics Playground

Unlock the full potential of your RNA-seq and proteomics data.

Try Omics Playground

About the Author

Axel Martinelli

Axel Martinelli’s academic background is in molecular biology and parasitology. He earned a Ph.D. on the genetics of strain-specific immunity against malaria infections and a master’s degree in bioinformatics with specialization in the analysis of omics data. During his postdoctoral career, he worked on genomics and transcriptomics studies and is currently the head of biology at Bigomics Analytics.