In previous version of RNAInter, we developed confidence score based on different detect method. To a certain extent, it can guide users to filter the RNA-associated interactions of interest. However, such a method was too arbitrary. In this version, the new confidence score was defined by integrating the trust of the scientific community(S), the trust of experimental evidence(E) and types of tissues/cells(T), which increases the scoring reliability.

1. The trust of the scientific community(S):

The trust of the scientific community can be reflected by the number of citations and publication years derived Google Scholar. A higher number of citations corresponds to a higher confidence score. This metrics S can be calculated as follow:

where i is the number of publications or prediction tools which can support the interaction, r_i represents the citations of the i-th publication or prediction tool, and y_i stands for the publication year of the i-th publication or prediction tool.

2. The trust of experimental evidence(E):

For the trust of experimental evidence, since a small-scale experiment is more reliable than a large-scale screening, publications describing few interactions contributed more than those describing many interactions. This metrics E can be calculated as follow:

where i is the number of publications or prediction tools which can support the interaction, n_i represents the interaction number described or predicted by the i-th publication or prediction tool.

3. The types of tissues/cells(T):

Additionally, the more tissues/cells the interaction was detected in, the higher confidence score the interaction has. This metrics T can be calculated as follow:

The three metrics were scaled in the range of [0, 1], separately. Then the Euclidean distance of them was calculated as the confidence score.

The final confidence scores were log2-transformed and scaled to [0, 1]. Accordingly, interactions reported by highly cited papers and detected in more tissues/cells will obtain a higher confidence score.

We evaluated new confidence scoring system based on the three different levels of supporting evidence. The interactions can be divided into three levels according to the different sources of evidence from which they were derived. Interactions with strong evidence were supported by strong experiments, while those with weak evidence by weak experiments and those with predicted evidence were only supported by predicted methods. The greatest enrichment score intervals of interactions with experimental evidence (blue and green bar) were from 0.2-0.3, while those of interactions with predicted evidence (red bar) were from 0.1-0.2. The mean scores of interactions with strong and weak evidences were 0.2886 and 0.2767, which was obviously higher than 0.1814, mean of interactions with predicted evidence.

This tutorial is as follows.

1. Firstly, we have to choose the type of keyword. There are three keyword types in our search as the picture shows. In this example, we choose RNA/Protein/DNA Symbol as the keyword type.

2. Next, we enter the keyword according to the keyword type selected in the previous step. In this example, we choose 'MEG3' as the keyword.

3. Then select the category for the keyword you entered. In this example, we choose 'lncRNA' as the category of the keyword 'MEG3'.

4. The next step is to choose the type of interaction. If you want to search for interaction of proteins with a particular RNA, you can choose 'RNA-Protein interaction'. In this example, we choose 'RNA-Protein interaction'. Under this condition, we can get RNA('MEG3')-protein interactions.

5. Then select the species for the keyword you entered. In this example, we choose 'Homo sapiens' as the species of the keyword 'MEG3'.

6. You can also choose the type of method that detects interaction as the filter. In this example, we want query the interaction detected by strong experimental evidence, so we choose 'Strong Experimental Evidence'.

7. We provide a score for each interaction. The greater the value, the higher the credibility. To filter low-confidence interactions, in this example, we choose the 0.2 as the minimum score and 1.0 as the maximum score.

After several seconds, the result will occur.

1. Your search conditions are in the head of the web page.

2. You can use the filter option to further screen search results by interactor, interactor category and species.

3. All the interactions are represented in th table format.

Firstly, you can get general information including RNAInter ID, confidence score, interaction type and predicted binding sites in the detail page.

Secondly, you can also get the basic information, homology interaction, target region information, evidence support and references of each entry.

Thridly, the annotation of RNA editing, localization, modification, structure and assication with disease also been provided.

Fourthly, the interaction network, tissue specific expression from GTEx database and the dynamic expression of interactors in spermatogenesis and haematopoietic stem cell of each entry has been represented in RNAInter.

1. Click Entrez ID/miRBase Accession/PubChem CID to see its basic description in NCBI Gene/miRBase/NCBI PubChem Compound database.

2. Category, UniProt, aliases, other ids and description of each interactor symbol.

RNA editing information from Lncediting, RADAR and DARNED is provided.

RNA localization information from RNALocate is provided, include symbol, subcellular localization, tissue or cell line, PMID. Click each subcellular localization can jump to RNALocate database.

RNA modification information from RMBase is provided, include modification positions, modification types and genomic contexts for each RNA symbol.Click each modification position can get more detail information in RMBase detail page.

RNA secondary structure by the prediction tool of RNA structure is provided. Select any transcript accession to see its secondary structure.

The annotation with disease association from MNDR v3.0 is shown.

The evidence support includes strong evidence, weak evidence, computaional prediction.

The reference includes pubmed ID, the source of database, tissue or cell line and description.

Interaction network for each interactor were provided.(only show the top 100 interactions of each interactor ranked by confidence score in our database ). Click each edge will redirect to corresponding detail page of interaction data.

Tissue specific expression from GTEx database and dynamic expression of interactors in spermatogenesis and haematopoietic stem cell of interactor(s) are provided.

Pearson correlation coefficients of each stage of spermatogenesis/HSC lineage commitment were provided.

Integration of source databases which use different interactors naming conventions is challenging. To ensure maximal connectivity of data, we transform each interactor name found in the input sources to the appropriate naming convention.

1. For miRNA, we use miRBase ID and miRBase Accession.

2. For compound, we use NCBI PubChem Compound symbol and CID.

3. For histone modification, we use ChIPBase symbol.

4. For others, we use official Gene Symbol and Entrez ID.

5. For species, we normalized organism names according to NCBI Taxonomy Database.

IntaRNA is a program for the fast and accurate prediction of interactions between two RNA molecules is provided.

1. Input RNA sequences or upload files with FASTA format.

2. Select and check the parameters.

The search results include target, query, position and energy.

PRIdictor is a tool for predicting protein-RNA interaction. You should input RNA and(or) protein sequence(s) as follow:

The search results include sequence, confidence and prediction sites.

DeepBind is a tool for predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning. You should input target sequence(s) or upload a file with FASTA format, and select query protein in left box. as follow:

The search results include target symbol, protein and its predict score. The results can be download.

Abbreviation	Full name
ac4C	N4-acetylcytidine
acp3U	3-(3-amino-3-carboxypropyl)uridine
Am	2'-O-methylguanosine
Ar(p)	2-O-ribosyladenosine (phosphate)
Cm	2'O-methylcytidine
cmnm5s2U	5-carboxymethylaminomethyl-2-thiouridine
cmnm5U	5-carboxymethylaminomethyluridine
D	dihydrouridine
f5C	5-formylcytidine
galQtRNA	galactosyl-queuosine
Gm	2'-O-methylguanosine
I	inosine
i6A	N6-isopentenyladenosine
m1A	1-methyladenosine
m1G	1-methylguanosine
m1I	1-methylinosine
m1Y	1-methylpseudouridine
m2,2G	N2,N2-dimethylguanosine
m2G	N2-methylguanosine
m3C	3-methylcytidine
m3U	3-methyluridine
m5C	5-methylcytidine
m5C	5-methylcytidine
m5U	5-methyluridine
m5Um	5,2'-O-dimethyluridine
m5Um	5,2'-O-dimethyluridine
m62A	N6,N6-dimethyladenosine
m6A	N6-methyladenosine
m7G	7-methylguanosine
mcm5s2U	5-methoxycarbonylmethyl-2-thiouridine
mcm5U	5-methoxycarbonylmethyluridine
ncm5U	5-carbamoylmethyluridine
o2yW	peroxywybutosine
QtRNA	queuosine
t6A	N6-threonylcarbamoyladenosine
Tm	2'-O-methylguanosine
tm5s2U	5-taurinomethyl-2-thiouridine
tm5U	5-taurinomethyluridine
Um	2'-O-methylguanosine
xA	unknown modified adenosine
xG	unknown modified guanosine
xU	unknown modified uridine
Y	pseudouridine
Ym	2'-O-methylpseudouridine
yW	wybutosine

The RNA-associated interactions are collected from different types of resources under one common framework, including experimental literature mining and computational prediction evidence. The experimental methods were divided into strong detection methods and weak detection methods by a manual assignment, depending on the nature and qualitative annotation of the experiment method.

Strong detection methods include:

3C	3C-qChIP	3D-DSL
3D-FISH	3RACE	3UTR indicator assay
3UTR reporter assay	4C	5RACE
Affinity technology	Ago2-IP	AGO-CLIP
AGO-IP	Allele-specific ChIP	ASO assay
ATPase assay	Beta-galactosidase activity assay	BiFC
Bio-plex assay	Biotin pull-down assay	BrdU incorporation assay
BSP assay	BS-PCR	Bulge-loop miRNA RT-PCR
CHART	CHART-MS	Chemosensitivity assay
ChIP	ChIP-3C	ChIP-PCR
ChIP-qPCR	ChIRP	ChIRP-MS
ChIRP-PCR	ChIRP-qPCR	ChOP
ChRIP	Chromatin accessibility assay	Chromatography technology
circRIP	Cleavage assay	CLIP
CLIP-qPCR	Co-FISH	Co-IP
Copurification	Cross-linking assay	DNA-FISH
DNase I footprinting	Dot-Blot assay	Drug assay
Drug efflux assay	Dual fluorescent reporter assay	Dual luciferase reporter assay
ELISA	EMSA	Enzyme assay
EPR	Filter binding assay	Filter trap assay
FISH	FISH-immuno	Fluorescence reporter assay
Footprinting	FRAP	Gel electrophoresis
Gel zymography	Hi-C	HPLC
HRR assay	HuR-IP	Hybrid-PCR
ICC	iDRiP	IFA
IFS	IHC	IIF
Immunoassay	Immunoblot	Indicator assay
Inhibition analysis	IP	ISH
ITC	Label transfer technique	LC/MS
LC-MS/MS	Luciferase reporter assay	Mass spectrometry
MeDIP	meRIP-qPCR	Microscopy
miR-Mask assay	miRNA assay	miRNA qPCR
miRNA RT-PCR	mRNA decay assay	MS2-RIP
MSP	MTS assay	MTT assay
Mutation analysis	NMR	Northern blot
Northern hybridization	PAGE	PCR
Primer extension assay	Probe interaction assay	Proximity ligation assay
pSILAC	Pull-down assay	qChIP
qPCR	qRT-PCR	RACE
RACE-PCR	RAKE analysis	RAP
RAP-MS	REMSA	Reporter assay
Rescue assay	RFLP	RIP
RIP-PCR	RIP-qPCR	RISC-Co-IP
RISC-IP	RLM-RACE	RNA blot
RNA chromatography	RNA Co-IP	RNA footprinting assay
RNA pull-down assay	RNA TRAP	RNA-ChIP
RNA-FISH	RNA-protein pulldown assay	RNAscope
RNase-ChIP	RPA	RQ-PCR
RT2-PCR	RTCA	RT-PCR
SDS-PAGE	Semi qPCR	Semi qRT-PCR
SILAC	smFISH	Southern blot
Stem-loop qPCR	Stem-loop qRT-PCR	Stem-loop RT-PCR
Strand specific RT-PCR	SYBR green PCR	SYBR green qPCR
TaqMan microRNA assay	TaqMan miRNA assay	TaqMan miRNA RT-PCR
TaqMan qPCR	TaqMan qRT-PCR	TaqMan RT-PCR
TLDA	Toeprinting assay	TRAP
Triplex capture assay	Two hybrid	U.V.-Crosslinking
Viral infectivity assay	Western blot	WST-1 assay
WST-8 assay	RISC-trap assay	UV-RIP
X-ray crystallography	Yeast two-hybrid analysis	Yeast three-hybrid analysis
WST	Immunostaining	MRM analysis
Transwell assay	TP-PCR	TaqMan assay
SPR assay	sqRT-PCR	Solid-phase assay
In situ hybridization	RNase-resistant assay	RNase protection assay
RNA-RNA pull-down assay	RNA-RNA in vitro interaction assay	RNA-RNA binding assay
RNA-protein filter binding assay	RNA-protein binding assay	RNA binding assay
RNA interference	RNA hybrid	RIP binding assay
Immunofluorescence assay	RNase H cleavage assay	Reverse transcription qPCR
Reverse RIP assay	Reciprocal IP	Real-time qPCR
Real-time RT-PCR	Protein precipitation assay	PAGE-Northern blot
MS2-TRAP	miRNA pull-down assay	In vitro binding assay
GST pull-down assay	EGFP reporter assay	ddPCR
DNA pull-down assay	CRISPR-Cas9	CISH
Biotin-coupled miRNA capture	Ago2-RIP	Ago2 pull-down assay

Weak detection methods include:

454 sequencing	Array	ATAC-seq
Bisulfite genomic-sequencing	BS-seq	CHART-seq
ChIP-chip	ChIP-seq	ChIRP-seq
ChOP-seq	ChRIP-seq	CLASH
CLEAR-CLIP	CLIP-seq	dCLIP
Deep sequencing	Degradome-seq	diMARGI
DNase-seq	eCLIP	FLASH
Genome-wide transcriptome sequencing	GMUCT	GRO-seq
High-throughput sequencing	HiSeq	HITS-CLIP
iCLIP	LIGR-seq	MARIO
MeDIP-seq	Microarray	miRNA array
miRNA-seq	NGS	PAR-CLIP
PARE	PARIS	pxMARGI
RACE-seq	RAP-seq	RIA-seq
RIP-Chip	RIPiT-seq	RIP-seq
RISC-seq	RNA CaptureSeq	RNA ChIP-on-chip
RNA-seq	sCLIP-seq	Sequencing
Small RNA Ultrahigh throughput sequencing	smRNA-seq	Solexa sequencing
SPLASH	sRNA-seq	TaqMan array
TaqMan microRNA array	TaqMan miRNA array	Tiling array
uvCLAP	miRMA PCR array	HuprotTM protoarray
Protein microarray	high-throughput protein-RNA interaction analysis	circRNA profiling analysis
circRNA microarray	4C-sequencing

Prediction methods include:

Bayesian target prediction algorithm	catRAPID	cRep
Cupid	Inparanoid	MicroInspector
miRanda	MirTarget	oRNAment
PARalyzer	pirScan	PITA
Rep	RNAhybrid	IntaRNA
TargetScan