LncACTdb, with the first version released in 2015, and the second version released in 2019, is a continually updated database that is used specifically to store experimentally validated ceRNA interactions manually curated from different species and disease states. Two versions of LncACTdb were both published on Nucleic Acids Research. Up to now the two papers have been cited 207 times (Google Scholar), and have been selected as ESI Highly Cited Papers in Web of Science. LncACTdb database is visited about 10,000 times from 73 countries(before Sep 2021). Emerging evidence suggests that microRNA (miRNAs) are themselves regulated by endogenous molecules carrying miRNA binding sites, such as lncRNAs, pseudogenes and circular RNAs. These competitive inhibitors bind miRNAs and competitively sequester them from their natural targets. The kind of competing endogenous RNAs (ceRNAs), act to dynamically buffer the expression of each other in different physiological and pathological processes. Previously, we developed LncACTdb database, which can serve as a tool for dissecting the ceRNA regulation in various cancers and identifying novel cancer biomarkers. Since its first release in 2015, more ceRNA interactions have been published, especially experimentally validated ceRNAs in various disease states, and a public resource of high-quality curated experimentally validated ceRNA interactions and personalized ceRNA networks remains unavailable. There is a great need to update LncACTdb with more resources and improved tools.

Figure 1. The development of LncACTdb database and collection of ceRNA related studies in recent years.

LncACTdb 3.0 is a comprehensive database of experimentally supported competing endogenous RNA (ceRNA) interactions and personalized networks contributing to precision medicine. LncACTdb 3.0 is freely available at http://bio-bigdata.hrbmu.edu.cn/LncACTdb or http://www.bio-bigdata.net/LncACTdb. We have updated LncACTdb 3.0 database with several new data and features, including: (i) more than 5,667 experimentally validated ceRNA interactions across 24 species and 536 diseases/phenotypes through manually curating of published literature; (ii) personalized ceRNA interactions and networks for 16,228 patients from 62 datasets in TCGA and GEO; (iii) manual curation of ceRNA sub-cellular and extracellular vesicle locations from literatures and related data sources; (iv) more than 10,000 experimentally supported lncRNA biomarkers associating with tumor diagnosis and therapy; (v) curation of lncRNA\mRNA\miRNA expression profiles with clinical and pathological information of thousands of cancer patients. Under the complex disease pathology and patient-specific background, a panel of improved tools have been developed to explore ceRNA effectson individuals. For example, the Network tool provides global view of lncRNA related ceRNA networks, patient-specific ceRNA networks, and customize designed ceRNA networks. Collectively, we believe that LncACTdb 3.0 will provide novel insights into the studying of disease pathology on individual level and further contribute to precision medicine.

Figure 2. The data expansion and featuresof LncACTdb 3.0.

In LncACTdb 3.0, we performed manual curation of experimentally supported lncRNA biomarkers to provide new insight into tumor diagnosis and therapy.We used the following combination of key words “(circulating OR drug-resistant OR prognostic OR immune OR metastasis OR recurrence OR cell growth OR EMT OR apoptosis OR autophagy) AND (lncRNA)” to collect diagnostic and therapeutic biomarkers. A biomarker was selected if the lncRNA wasassociating with this processes by over-expression, RNA knockdown or other functional experiments. Finally, a total of 10,084 experimentally supported lncRNA biomarkers associated with drug resistance, circulation, survival, immunity, metastasis, recurrence, cell growth, EMT, apoptosis, and autophagy were manually curated from literature, and integrated into the LncACTdb 3.0 database.

First,candidate ceRNA pairs were collected from two databases: starBase v2.0(9) and LncACTdb 2.0(6). A total of 108,668 candidate ceRNA regulations were collected. To verify whether these ceRNA pairs were associated with each other in a specificsample we used a published method for sample-specific network construction based on probability theory to identify ceRNA networks for the sample(10). We assume that each ceRNA pair may have an association in some samples but not in other samples due to differences in sample types.

Figure 3. Sample-specific ceRNA networks construction and our statistical model.

We determined whether lncRNAs and mRNAs were related in a specific sample by testing the statistical independence of the candidate ceRNA expression values in the same sample. For a ceRNA pair of x(mRNA) and y(lncRNA) in sample q, we calculated the following statistic:

Where n is the total number of samples. n_x^qand n_y^q are predetermined integers. We set n_x^q= n_y^q = 0.1n . We draw the first two boxes near n_x^q and n_y^q , based on the predetermined n_x^q and n_y^q , and then we have the third box, which is simply the intersection of the previous two boxes. Thus, we can obtain the value of n^xy_q by counting the plots in the third box. If x and y are independent of each other, this statistic follows a standard normal distribution and the mean value and variance for the n samples are 0 and 1, respectively. Therefore, we can determine the significance of the x, y correlation with this statistic. edge_xy^q is set to 1 in the network of sample q with a false discovery rate (FDR) < 0.05. We retained pairs that meet FDR <0.05 for network construction in a specific sample. The algorithm requires that expression data set must include both mRNA and lncRNA expression and it works better when the number of samples is greater than 100.This method is not sensitive to the normalisation method for gene expression matrices. In RNA-seq data, the statistic may result in zero due to experimental errors, which is meaningless in biology and may produce errors in the data analysis. Hence, we treat the zeros in the following way (10): (1) If we cannot distinguish whether or not the zeros result from zero expression or the experimental errors, edge_xy^q is set to 0 when x_q = 0 or y_q = 0 without the consideration of the statistic. (2) If we know that the zeros result from the zero expression, edge_xy^q is determined by the statistic.

Based on the above pipeline, we purified ceRNA pairs in a cancer-specific way. The specificity of the ceRNAs was characterized quantitatively by calculating the specificity score using a previously described method (11-13). For a ceRNA, the specificity score was calculated as:

Where N is the number of cancers and x_i is the percentage of samples in which the ceRNA can be found in a cancer. The value of x_i is normalized to the maximum percentage values of cancers. For example, the specificity score for a ceRNA with percentage profile of ‘0 0 0.2 0 0 0 0 0.2 0 0 0.8 0’ is calculated as 0.9545. The range of specificity scores is between 0 and 1, while a perfect specific pattern would be scored as 1.According to an earlier study(1), an specificity score<0.15 indicates a housekeeping gene. We used the specificity score>0.5 as threshold to purified cancer-specific ceRNAs.

In LncACTdb 3.0, the Network tool constructs and illustrates ceRNA network in the following ways:

(i) Gene-centric
For a lncRNA or mRNA, LncACTdb 3.0 provides a global view of all possible related ceRNAs relationships. A network consisting of this lncRNA or mRNA and its associated competing neighbors was constructed and illustrated by a Java script plugin ECharts (V4.0). The ceRNA network scale can be reset by selecting different steps of neighbours. In the one-step-neighbours scale, the top 20 competing mRNA partners (ordered by activity score) of the lncRNA were illustrated. In the two-step-neighbours and three-step-neighbours scale, this network will expand to another 20 and 40 competing lncRNAs and mRNAs.

(ii) Patient-centric
For a cancer sample, LncACTdb 3.0 provides a patient-specific ceRNA network. In this section, users can input the sample name to obtain a patient-specific ceRNA network. In the network, ceRNA interactions were determined to be specifically active a sample by testing the statistical independence of the candidate ceRNA expression values. When users move the cursor over a node in the network, all edges and nodes connected to it are highlighted. Different network layouts, such as the circular layout and force layout can be used to illustrate the network.

(iii) User-designed
For user selected data, LncACTdb 3.0 provides a user-designed ceRNA network in which lncRNAs, miRNAs, mRNAs and diseases were determined by users. This network were constructed based on the experimentally validated ceRNA relations in LncACTdb 3.0. The network degree of ceRNAs was list in a data table. Users can starts a new search of experimentally validated ceRNA relations by clicking an interesting gene in the network.

1* Kanehisa, M., Goto, S., Sato, Y., Furumichi, M. and Tanabe, M. (2012) KEGG for integration and interpretation of large-scale molecular data sets. Nucleic acids research, 40, D109-114.

2* Croft, D., Mundo, A.F., Haw, R., Milacic, M., Weiser, J., Wu, G., Caudy, M., Garapati, P., Gillespie, M., Kamdar, M.R. et al. (2014) The Reactome pathway knowledgebase. Nucleic acids research, 42, D472-477.

3* Schaefer, C.F., Anthony, K., Krupa, S., Buchoff, J., Day, M., Hannay, T. and Buetow, K.H. (2009) PID: the Pathway Interaction Database. Nucleic acids research, 37, D674-679.

4* Liberzon, A., Birger, C., Thorvaldsdottir, H., Ghandi, M., Mesirov, J.P. and Tamayo, P. (2015) The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell systems, 1, 417-425.

5* Hanahan, D. and Weinberg, R.A. (2011) Hallmarks of cancer: the next generation. Cell, 144, 646-674.

6* Wang, P., Li, X., Gao, Y., Guo, Q., Wang, Y., Fang, Y., Ma, X., Zhi, H., Zhou, D., Shen, W. et al. (2019) LncACTdb 2.0: an updated database of experimentally supported ceRNA interactions curated from low- and high-throughput experiments. Nucleic Acids Res, 47, D121-D127.

7* Yuan, H., Yan, M., Zhang, G., Liu, W., Deng, C., Liao, G., Xu, L., Luo, T., Yan, H., Long, Z. et al. (2019) CancerSEA: a cancer single-cell state atlas. Nucleic Acids Res, 47, D900-d908.

8* Hanzelmann, S., Castelo, R. and Guinney, J. (2013) GSVA: gene set variation analysis for microarray and RNA-seq data. BMC bioinformatics, 14, 7.

9* Li, J.H., Liu, S., Zhou, H., Qu, L.H. and Yang, J.H. (2014) starBase v2.0: decoding miRNA-ceRNA, miRNA-ncRNA and protein-RNA interaction networks from large-scale CLIP-Seq data. Nucleic Acids Res, 42, D92-97.

10* Dai, H., Li, L., Zeng, T. and Chen, L. (2019) Cell-specific network constructed by single-cell RNA sequencing data. Nucleic Acids Res, 47, e62.

11* Yanai, I., Benjamin, H., Shmoish, M., Chalifa-Caspi, V., Shklar, M., Ophir, R., Bar-Even, A., Horn-Saban, S., Safran, M., Domany, E. et al. (2005) Genome-wide midrange transcription profiles reveal expression level relationships in human tissue specification. Bioinformatics, 21, 650-659.

12* Wang, P., Ning, S., Zhang, Y., Li, R., Ye, J., Zhao, Z., Zhi, H., Wang, T., Guo, Z. and Li, X. (2015) Identification of lncRNA-associated competing triplets reveals global patterns and prognostic markers for cancer. Nucleic Acids Res, 43, 3478-3489.

13* Guo, Q., Wang, J., Gao, Y., Li, X., Hao, Y., Ning, S. and Wang, P. (2020) Dynamic TF-lncRNA Regulatory Networks Revealed Prognostic Signatures in the Development of Ovarian Cancer. Front Bioeng Biotechnol, 8, 460.

A. Enter the name/ID of mRNA/LncRNA.

B. Enter ceRNA and disease information and drag the input field below to select the cancer specificity threshold for ceRNA.

C. This column shows the cancer specificity score for the same row of ceRNA.

A. Select the data set and set the signal pathway score.

B. View sample details.

C. Total number of ceRNAs possessed by the sample and the pathway score.

A. All disease datasets included in LncACTdb 3.0.

B. The ceRNA network diagram in this dataset.

A. All biomarker information contained in LncACTdb3.0.

B. Details of this biomarker.

A. Click and select 5/10/20/30 top enriched functions.

B. The enriched pathways of hit downstream genes. / The enriched GO terms of hit downstream genes.

A. The COX regression analysis result of the hit completing triplet.

B. Survival curves based on mean/median expression and forest plot.

A. CeRNA networks in different samples.

B. The hit lncRNA-cenrtic ceRNA network.

C. Experimental confirmation of the ceRNA network obtained from the data.

A global view of sample states/behaviors such as angiogenesis, apoptosis, cell cycle, invasion, proliferation, stemness etc.

A. Enter the sample ID

B. 10 sample state scores

C. Tumor cell score

A. LncACTdb 3.0 and hot words from previous work. (Changes over time) .

A. Click the Detail button for more information .

A. Please describe your problem in detail .

B. Our contact information and address.