Publicly available Total RNA and small RNA Transcriptome Sequencing data from Human and mouse B and T cell developmental stages and T-ALL and CLL Leukemia data (Table 1).

Raw reads from the small RNA sequencing datasets were checked for their quality using the FastQC(https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Based on phred score quality cut-off, low quality reads were trimmed from both the ends keeping a minimum of 18 nucleotides from an initial read length of 26-36 nucleotides. Post trimming the reads were mapped to the mouse reference (mm10) and human reference (hg38) genome, respectively using the Bowtie2 (38)aligner. The resulting SAM files were converted to compressed BAM format using the SAMtools package (39). In order to quantify the expression, the read counts per miRNA transcript were obtained using the BEDtools suite (40). The read counts were normalized by using the RPM method, i.e., RPM = number of reads mapped per transcript/(total number of mapped reads/10⁶). The rpm values were extracted and analyzed for difference in expression. The miRNA known to regulate RAG1 and RAG2 were obtained from miRDB (41) and TargetScan (42). The miRNAs targeting RAGs were obtained based on a custom prediction available in the databases based on the presence of seed sequences that bind to the 3’UTR of RAG1 and RAG2. The shortlisted miRNA were then checked for their expression in the small-RNA datasets (Figure S1A).

For the T-ALL and CLL patients’ datasets, Normalized Read count files for miRNA transcripts and mRNA transcripts from 15 CLL patients were downloaded from the GEO datasets mentioned in Table 1. Raw read count files for miRNA and mRNA transcripts for 48 T-ALL patients were downloaded from the GEO database. RPM values for miRNA and RPKM values for mRNA were calculated for both the datasets using the following formulae, RPM = read count/(total number of reads mapped/10⁶) and RPKM = read count/(transcript length/1000) ∗(total number of reads mapped/10⁶). The normalized values for RAG1, its interacting partners and the miRNA known to regulate RAG1 were plotted in a heatmap using the pheatmap package in R.

The database has been structured and constructed using the Shiny package from RStudio^© (https://shiny.rstudio.com/). We are providing an offline and online version of the database; the online version is deployed on the Rstudio server (https://www.shinyapps.io/) which works on static TSV/CSV files resulting from the analysis which contains expression data. At the same time, the offline version works on SQLite database. The data is fetched and rendered in a table or graphical format on the UI based on the user query.

The application is written in 3 separate R files, namely ui.R, server.R and global.R. The ui.R renders user interface, whereas server.R does processing of data as per user input and global.R reads all CSV files required for the database (Figure S1B). List of packages used to read in and visualize the data are mentioned in Figure S1C.

The database URL lands on the “Home” page. On the left side there is a navigation panel that lets the user choose the next page they want (Figure 1A). This page displays information about RAG1, RAG2 and its genomic location on chr11, and its role in V(D)J recombination in B and T cells as part of the RAG nuclease complex. This page offers an option for the users to be able to navigate to the RAG1 related pages of HGNC, Entrez Gene, Ensembl, OMIM, UniProtKB and Atlas of Genetics and Cytogenetics in Oncology and Haematology databases along with a link to a diagram describing RAG mechanism through the “Mechanism of Action” tag (Figure 1B).

Based on a previous study from the lab, where the role of miRNA in the regulation of RAGs was investigated, small RNA-seq datasets consisting of mouse developmental stages of B and T cells were downloaded. Using a standard transcriptome analysis pipeline with mm10 as a reference genome for alignment, followed by calculation of normalized expression, miRNA regulating RAG1 and RAG2 were analyzed for the difference in expression across developmental stages of normal mouse B and T cells. It is known that expression of RAGs differs in a stage-dependent manner and hence, the miRNA expression data were analyzed for a corresponding inverse expression profile of miRNA. Transcriptome datasets from normal B and T cells were also obtained and analyzed (Table 1). These expression profiles are displayed on the mouse page of miRAGDB.

The page begins with a tabular display of RAG regulating miRNA expression values in normal T cell development stages on the left alongside a corresponding heatmap representing the same expression profiles on the right (Figures 2A, B). Following the miRNA expression, the page moves to tabulated RAG1 expression values in normal T-cell stages followed by a bar graph representing the same. In the end, just as it was displayed for T- cells, there is a side-by-side display of RAG1 and RAG2 regulating miRNA expression values in B cell development stages on the left and the corresponding heatmap on right.

Datasets consisting of small-RNA sequencing across human B cell development stages were downloaded (Table 1) and processed using hg19 as the human reference genome. Similar to the mouse page, this area displays miRNA expression profiles in tabular format along with a visualization in the heatmap and bar graph forms. The difference is that this page is rather interactive. In the beginning, it starts in the same way as the mouse page with miRNA expression values across stages of B cell development in a table on the left and a corresponding heatmap on the right (Figures 3A, B). Going forward, we present the same miRNA B cell data in the form of individual bar graphs available to the users as per their selection from a dropdown list of the miRNAs. Further, we present a fascinating utility where from a table of expression profiles of miRNA across B-cell development stages, the user can choose any number of miRNAs (at least two) in any order, and a corresponding heatmap representing the expression profile will be displayed on the right in that specific order (Figures 3C, D). However, the shade of the color corresponding to the expression value in the heatmap for every miRNA for each stage will differ slightly with the addition or removal of every miRNA. At the end just as in the mouse page, the human page displays expression levels of miRNA against RAG2 in Human B cell development stages.

The following two pages deal with the expression of RAG1 and the miRNA regulating it in two datasets, Chronic lymphocytic leukemia, and T-cell Acute Lymphoblastic Leukemia, respectively.

Both the pages begin by displaying an expression table of RAG1 regulating miRNA across 15 CLL patient samples (17 miRNA spread across two pages) and 48 T-ALL samples. This table includes a search dialogue box where the user can search for one of the 17 miRNAs and obtain expression values specific to that miRNA. Further down is a heatmap of those miRNAs (Figures 4A, B) and at the end is a Bar graph showing RAG1 expression across the 15 CLL patients and 48 T-ALL patients (Figures 4C, D).

Based on the literature study, we prepared a list of genes known to regulate and interact with RAG1 (POU2F2, E2A, FOXP1, MYB, BCL11A, FOXO1, EBF1, IKAROS, and PAX5) and RAG2 (BCL11A, E2A, PAX5, SATB1, CDK2, GATA3, IKAROS, RUNX1 and ETV6). We identified the expression profiles of these genes in human and mouse B cell development data and T cell development data. The RAG regulators page displays these expression values in a tabular format followed by a heatmap showing the same (Figures 5A, B). After that, the expression levels of genes known for RAG1 degradation and localization in B and T cell development data has been shown in a tabular as well as heatmap format. The next section shows expression data of RAG2 regulators in B and T cell development data across their development stages. The page ends with Expression of RAG regulators in CLL condition.

The following two pages are curations of data from other resources with information specific to RAG1 and RAG2. Since a significant highlight of this database is miRNA involved in RAG1 and RAG2, on the next page, we have cataloged all the circular RNAs that are known to make a sponge with every miRNA that we have mentioned in the previous pages. This instance presents the user with a drop-down list of miRNAs to choose from, resulting in a list of circRNA and its coordinates on the human genome, which might act as a sponge. Every circRNA has a corresponding UCSC genome browser link attached to it that displays the chromosomal location of the circRNA(Figure 6A). Also, an MGI page of RAG1 with 158 mouse RAG1 MGI hyperlinks to the expression of RAG1 in various tissues. The table consists of the RAG1 MG1 ID, the genes symbol “Rag1”, the link to the corresponding MGI database, the Assay type that determined the expression, the development stage of the mice at which the data was extracted, the tissue followed by the image id of the tissue expression data (Figure 6B).

The following two utilities are highly user-friendly. The first is a Download section where all the data displayed on the database can be downloaded by the user with three available options, mouse data, human data, or all the files together (Figure 6C). One of the most exciting utilities comes next: the page where users can upload their own data and visualize it by adhering to the following steps. 1) It starts by asking you to upload a file of your choice in a tabular format by clicking on the “Browse button”, 2) If your input file has a header, you check the “Header” box followed by 3) choosing the separator that is present in your file (Comma, Semicolon or Tab) (Figure 6D). Once all these three steps are followed, the values of your file will be displayed in a tabular format followed by a heatmap depicting the same.

Using the database, miRNAs against RAG regulators can be identified, and their significance in regulating genomic integrity be studied. As an example, we would like to draw focus on miR-29c-3p. When gathering and analyzing expression data of miRNAs known to regulate RAG1 across B cell developmental stages for the database, we came across miR-29c-3p. We observed that the expression of this miRNA is negligible in the preB and proB cell stages, while it increases drastically in the mature stages of B cells, indicating an inverse pattern of expression. These findings were validated experimentally using knockdown and overexpression studies. It was also seen that in 87% of CLL patients and 79% of T-ALL patients, the expression levels of miR-29c-3p and RAG1 were inversely correlated (28).