Long noncoding RNAs (lncRNA) have emerged as promising biomarkers in cancer diagnosis, treatment, and prognosis. Recent studies suggest that a large number of coding gene expression microarray probes could be reannotated as lncRNAs. Microarray, once the most cutting-edge high-throughput gene expression technology, has been used for thousands of cancer studies and has brought invaluable resources for studying the functions of lncRNA in cancer development. However, a comprehensive lncRNA resource based on microarray data is still lacking. Here, we present lnCAR (lncRNAs from cancer arrays), a comprehensive open resource for providing expression profiles and prognostic landscape of lncRNAs derived from reannotation of public microarray data. Currently, lnCAR contains 52,300 samples for differential expression analysis and 12,883 samples for survival analysis from 10 cancer types. lnCAR allows users to interactively explore any annotated or novel lncRNAs. We believe lnCAR will serve as a valuable resource for the community focused on lncRNA research in cancer.

Significance:

lnCAR, a new interactive tool of reannotated public cancer-related microarray data, provides expression profiles and prognostic landscapes of lncRNAs across thousands of samples and multiple cancer types.

Long noncoding RNAs (lncRNA), defined as transcripts longer than 200 nucleotides that do not encode proteins, have gained widespread attention in recent years for their important roles in cell growth, differentiation, apoptosis, and genomic imprinting (1–4). Accumulating studies have revealed that lncRNAs play important roles in the tumorigenesis of most cancers. For example, lncRNA HOTAIR has been identified as an oncogene gene in many cancers (5), whereas MEG3 has been proposed as a tumor suppressor (6).

Advances in high-throughput technology, such as microarrays and next-generation RNA sequencing (RNA-seq) have resulted in large amounts of expression data across different cancers. However, the lower expression levels have made it challenging to obtain lncRNA information. Furthermore, to prioritize lncRNAs of important function and potential clinical relevance, it's urgent to evaluate the associations of lncRNA with RNA expression and clinical variables (e.g., tumor status, survival information). To fill these needs, many systematically annotated lncRNA repositories have been generated. Some of them have been developed for investigating the relevance of lncRNA expression to cancers based on RNA-seq data, such as MiTranscriptome (7) and TANRIC (8). Some databases were designed to collect cancer-related lncRNAs identified by low-throughput experiments (e.g., real-time PCR and Northern blot), such as Lnc2Cancer (9), LncRNADisease (10), EVLncRNAs (11), and MNDR v2.0 (12). Nevertheless, microarrays, the most abundant data pool for cancer research (Supplementary Fig. S1), have not been systematically mined for lncRNA expression.

Previous studies have suggested that a large number of probes can be reannotated for investigating lncRNA expression (13, 14). These microarrays are invaluable resources for lncRNA research in cancer, but a comprehensive lncRNA resource based on microarray data is still lacking. In this study, we present lnCAR (lncRNAs from cancer arrays; Fig. 1), a comprehensive open resource for the exploration of lncRNAs obtained by reannotating the gene expression microarray data. Using lnCAR, users are able to rapidly and intuitively explore differentially expressed lncRNAs of interest for various types of contrast experiments, such as “tumor versus normal” and “drug response versus non-response”. In addition, lnCAR is also a powerful platform for querying associations between lncRNA expression and survival outcome.

Figure 1.

Flowchart of database construction.

Figure 1.

Flowchart of database construction.

Close modal

Data sources

We developed a text-mining pipeline to automatically extract the sample and clinical information of the gene expression studies for the cancers of interest from Gene Expression Omnibus (GEO) database. Briefly, study IDs (GSE accessions) were obtained by providing cancer-related key words (Supplementary Table S1) along with the “filters: Expression profiling by array” (15). The description file in “soft” format for each study was parsed to obtain the sample list GSM (GEO Sample) IDs. To avoid sample duplicates, unique GSM IDs were retained. Then, the description file in “soft” format for each sample was parsed to extract the sample and clinical information such as tissue type (tumor or normal), grade, stage, metastasis, treatment, survival, etc. To ensure the data quality, we manually curated all of data collected from the text-mining pipeline. The probe sequences of each microarray platform were downloaded from the GEO database or the websites of their corresponding companies. Some platforms and their related samples were removed when the probe sequences could not be downloaded because of a permission issue or expired resource. All of the expression datasets were collected before August 2017.

Probe reannotation and expression normalization

To reannotate the probe sequence, we first integrated the gene annotations from the Ensembl and RefSeq database (Supplementary Fig. S2). Then, the probe sequences from the different platforms were aligned to human genome and mapped to coding genes or lncRNAs according to their genomic coordinates (Supplementary Fig. S2; Supplementary Materials and Methods). The expression dataset from different platforms were normalized using different strategies. For the expression dataset from Affymetrix, raw CEL files were obtained when possible, and were normalized with the RMA algorithm (“affy” package v. 1.26 or “apt-probeset-summarize” in Affymetrix Power Tools v. 1.19.0; refs. 16, 17). For the dataset from Agilent, Illumina, and other platforms, the expression raw data were quantile-normalized using the Limma package (18). For the studies lacking raw data, we directly used the data matrix provided in the GEO database. All the data are normalized at probe level and log2-transformed. Next, the probes were mapped to gene names based on the probe reannotation file. The mean expression was used for the genes that are targeted by multiple probes.

Differential expression analysis

We categorized differential expression with one of the following conditions: cancer versus normal controls, high stage versus low stage, high grade versus low grade, metastasis versus primary, drug-treatment versus control, and “other features”. For each condition, the differential expression analysis was performed with the Limma package and a robust rank aggregation algorithm was utilized to integrate the lncRNA profiles in an unbiased manner (18, 19). The aggregation rank score (AR score) represents the integrated rank from the meta-analysis of the fold-change in the different microarray studies. (see Supplementary Materials and Methods for more details).

Survival analysis

All of the data we collected contained four types of survival data: overall survival, relapse-free survival, metastasis-free survival, and progression-free survival. We accessed the associations between lncRNA expression and survival via a univariate Cox regression analysis (20). To combine the prognostic data from the different studies, we performed a meta-analysis of the Z scores derived from the univariate Cox regression analysis. (see Supplementary Materials and Methods for more details).

Advanced annotations and other analysis

We provided advanced annotations of lncRNAs, such as the transcript information, structure, coding potential, and conservative score. In addition, cancer-related lncRNAs validated by low-throughput experiments were integrated as validation datasets. Moreover, our database also included coexpression analysis, pathway enrichment analysis, and competing endogenous RNA (ceRNA) analysis modules to help researchers further explore the function of interested lncRNAs. (see Supplementary Materials and Methods for more details).

Determination of cancer subtypes by lncRNA expression

The Affymetrix expression datasets with known cancer subtypes of colorectal cancer (n = 434), bladder cancer (n = 209), and breast cancer (n = 1,381) were used to identify the lncRNA expression subtypes (Supplementary Table S2). For each cancer, we used an unsupervised clustering algorithm with iterative adjustment that was modified from a previous study (21) to divide the tumor samples into different subtypes (Supplementary Materials and Methods). Next, a χ2 test was applied to investigate the correlation between the lncRNA expression subtypes and known cancer subtypes. For 616 patients with breast cancer with clinical data (Supplementary Table S2), we applied log-rank test to check the association between lncRNA subtypes or known subtypes and relapse-free survival. To explore the molecular characteristics of each subtype in breast cancer, we identified the lncRNA signatures by differential expression analysis using a moderated F test and performed the pathway enrichment analysis of protein-coding genes proximal to these lncRNA signatures by clusterProfiler package (see Supplementary Materials and Methods for more details; ref. 22).

Web interface implementation

All the analysis results and sample information were managed using MySQL tables. The web interfaces were implemented using Hyper Text Markup Language (HTML), Cascading Style Sheets (CSS), and JavaScript (JS). Furthermore, we allowed the genomic coordinates of the regions of interest as an input to search the expression and prognosis of any user-defined lncRNA (Supplementary Materials and Methods).

Detailed methods are available in the Supplementary Materials and Methods.

Data Summary

Gene reference.

As illustrated in Supplementary Fig. S3A, we obtained a comprehensive reference set composed of 19,805 coding genes and 28,240 lncRNAs.

Probe reannotation.

Among 107 gene expression microarray platforms from all available probe sequences, there were 48 platforms from Affymetrix, 32 from Agilent, 12 from Illumina, and 15 from other synthesis platforms. In total, 464,502 lncRNA probes representing 26,464 lncRNAs were obtained (Supplementary Fig. S3B; Supplementary Table S3).

Cancer arrays.

In the current release, lnCAR contains 732 carefully curated expression microarray datasets covering 54,893 samples across 10 cancer types, that is, breast cancer, bladder cancer, cervical cancer, colorectal cancer, esophageal cancer, gastric cancer, liver cancer, lung cancer, and prostate cancer (Table 1). Among these samples, 52,300 had more than seven different clinic subtyping information sets, such as “tumor versus normal”, “high grade versus low grade,” and “treatment versus control” (Supplementary Table S4), whereas 12,883 had survival information such as “overall survival” and “relapse-free survival” (Supplementary Table S5). In addition, 566 cancer-related lncRNAs generated from low-throughput technology were manually curated as validation data (Supplementary Table S6). To the best of our knowledge, lnCAR represents the largest public resource for lncRNA research.

Table 1.

Summary of the data resources of the current lnCAR release

Bladder cancerBreast cancerCervical cancerColorectal cancerEsophageal cancerGastric cancerLiver cancerLung cancerOvarian cancerProstatic cancer
 1,912 17,421 1,172 8,404 1,370 3,082 4,467 5,292 4,554 4,636 
Tumor vs. normal 606 3,970 532 5,128 1,168 1,919 3,476 3,795 2,157 2,653 
High grade vs. low grade 1,068 9,205 29 771 82 277 42 598 2,686 1,953 
High stage vs. low stage 440 2,710 457 3,069 292 1,330 526 1,141 2,488 351 
Metastasis vs. primary 181 1,073 28 100 183 
Treatment vs. control 862 293 314 30 191 611 38 177 344 
Other features 10,276 197 1,008 193 43 349 1,649 
 418 6,148 300 1,675 179 792 196 1,906 1,188 81 
Overall survival 325 2,570 1,179 179 83 196 1,853 1,188 
Metastasis-free survival 1,663 
Relapse-free survival 93 4,383 300 1,121 709 108 720 506 81 
Progression-free survival 226 53 79 
Bladder cancerBreast cancerCervical cancerColorectal cancerEsophageal cancerGastric cancerLiver cancerLung cancerOvarian cancerProstatic cancer
 1,912 17,421 1,172 8,404 1,370 3,082 4,467 5,292 4,554 4,636 
Tumor vs. normal 606 3,970 532 5,128 1,168 1,919 3,476 3,795 2,157 2,653 
High grade vs. low grade 1,068 9,205 29 771 82 277 42 598 2,686 1,953 
High stage vs. low stage 440 2,710 457 3,069 292 1,330 526 1,141 2,488 351 
Metastasis vs. primary 181 1,073 28 100 183 
Treatment vs. control 862 293 314 30 191 611 38 177 344 
Other features 10,276 197 1,008 193 43 349 1,649 
 418 6,148 300 1,675 179 792 196 1,906 1,188 81 
Overall survival 325 2,570 1,179 179 83 196 1,853 1,188 
Metastasis-free survival 1,663 
Relapse-free survival 93 4,383 300 1,121 709 108 720 506 81 
Progression-free survival 226 53 79 

NOTE: Bold data is the sum of the data resources.

Web interface and usage

lnCAR provides a user friendly web interface that contains four modules: Explore, My lncRNA, Statistics, and Download (see Supplementary Video S1).

Explore.

lnCAR provides two analysis modes to explore lncRNAs of interest: differential expression analysis and survival analysis. In the “differential expression” mode, an interactive heatmap was implemented to present the meta-score from differential expression meta-analysis for comparisons across cancer subtypes, such as “tumor versus normal” and “high grade versus low grade” (Supplementary Fig. S4A). Users are allowed to explore the differential expression of lncRNAs across cancer subtypes by searching and sorting functions. A selection box to quickly locate cancers and a text box to search lncRNAs of interest are provided. Furthermore, users can sort the lncRNAs by clicking the column name. The detailed information for the differential expression of their interested lncRNA in a certain cancer was shown when clicking the “meta-score” box (Supplementary Fig. S4B–S4D). Some visualized figures, such as a gene expression boxplot, an mRNA–lncRNA expression network, a bubble chart of the pathway, and a circos plot of the ceRNA network, are generated in real time to allow investigation of the potential effects of lncRNAs (Supplementary Fig. S4E). In addition, the verification result obtained by using the low-throughput experiments was provided (Supplementary Fig. S4F). Similarly, the “survival” mode offers an overview of the prognostic impact of lncRNAs by a meta-score heatmap (Supplementary Fig. S5A). Users can explore the relationship between lncRNAs and different survival outcomes, such as overall survival and relapse-free survival across cancer types (Supplementary Fig. S5B).

My lncRNA.

lnCAR allows users to explore their user-defined lncRNAs by searching genomic regions of the lncRNAs. For example, by searching the genomic region “chr12:53,962,308-53,974956″ in breast cancer (Supplementary Fig. S6A), users could obtain the results of differential expression analysis (Supplementary Fig. S6B) and survival analysis (Supplementary Fig. S6C) for the lncRNA probes located in this region.

Statistics and download.

Detailed statistics on the data in lnCAR were provided in the “Statistics” module (Supplementary Fig. S7A–S7D). And users can download all analysis results from the “Download” module.

Oncogenic role of HOTAIR in breast cancer revealed by lnCAR

HOTAIR, known as HOX transcript antisense RNA, is transcribed from the antisense stand of the HOXC gene cluster and involved in the epigenetic regulation of HOXD genes by interacting with polycomb repressive complex 2 (PRC2; ref. 23). HOTAIR is highly expressed in different tumors, such as breast, gastric, and cervical cancer, and may play important role in the process of cancer diagnosis and treatment (5). To demonstrate the potential value of discovering cancer-related lncRNA for lnCAR, we explored the expression and clinical association of HOTAIR in breast cancer using lnCAR. In line with previous studies, the differential expression analyses show significant upregulation of HOTAIR in tumor samples compared with normal samples in nine of 23 GEO datasets, which is also confirmed in The Cancer Genome Atlas (TCGA) dataset (Fig. 2A). Besides, increased HOTAIR expression is related to advanced tumor stage and higher tumor grade (Supplementary Fig. S8A). In addition, HOTAIR is overexpressed in HER2-positive, estrogen receptor (ER)-negative, and progesterone receptor (PR)-negative patients (Supplementary Fig. S8B), which is consistent with the previous research (24). To predict the molecular mechanism of HOTAIR in breast cancer, we investigated the coexpression networks of HOTAIR using coexpression function that was implemented in lnCAR. As a result, the genes in HOXC cluster are highly coexpressed with HOTAIR in lnCAR and TCGA, consistent with previous studies (Fig. 2B). Consistent with previous studies, the survival analysis shows no significant association between the expression of HOTAIR and survival in lnCAR and TCGA (Fig. 2C), indicating that HOTAIR might not serve as an independent prognostic marker in breast cancer (25).

Figure 2.

The expression and survival analysis of lncRNA HOTAIR in breast cancer. A, The differential expression analyses form different studies in lnCAR and TCGA dataset indicate HOTAIR is significantly upregulated among tumor samples compared with normal samples. B, Coexpression analysis using lnCAR and TCGA datasets reveals HOXC cluster are highly coexpressed with HOTAIR. C, No significant association is identified between the expression of HOTAIR and clinical outcomes in lnCAR and TCGA datasets.

Figure 2.

The expression and survival analysis of lncRNA HOTAIR in breast cancer. A, The differential expression analyses form different studies in lnCAR and TCGA dataset indicate HOTAIR is significantly upregulated among tumor samples compared with normal samples. B, Coexpression analysis using lnCAR and TCGA datasets reveals HOXC cluster are highly coexpressed with HOTAIR. C, No significant association is identified between the expression of HOTAIR and clinical outcomes in lnCAR and TCGA datasets.

Close modal

A large collection of cancer-relevant lncRNAs in lnCAR

To further demonstrate the usefulness of the tool to discover new biology, we systematically investigated the potential clinical significance of lncRNAs across cancer types based on lnCAR resource. First, we found a large number of lncRNAs are significantly differentially expressed between tumor samples and normal samples (fold change ≥ 1.5 and Padj < 0.05), among which, colorectal cancer have the largest number of lncRNAs, followed by gastric cancer and lung cancer (Fig. 3A). A total of 2,075 lncRNAs are differentially expressed in more than six cancer types, suggesting broad pancancer functional importance of these lncRNAs (Fig. 3B). The top 10 upregulated and downregulated lncRNAs from the meta-analysis across cancer types are shown in Fig. 3C. Interestingly, most of them are associated with cancers in previous studies (Supplementary Table S7). For instance, Zhao and colleagues found lncRNA SNHG17 was significantly increased in breast cancer (26). A recent study reported that SNHG17 is markedly upregulated in colorectal cancer, and the overexpression of SNHG17 is correlated with tumor size, advanced tumor–node–metastasis (TNM stage), and lymphatic nodes metastasis (27). We found the expression of HAND2-AS1 was decreased in various cancers, including liver cancer and endometrial cancer. Consistent with this, previous studies reported that HAND2-AS1 is downregulated in many cancer types and that the downregulation of HAND2-AS1 is associated with tumor grade, lymphatic nodes metastasis, and relapse (28–30).

Figure 3.

A large number of differentially expressed lncRNAs in various cancer types. A, Bar plot showing the numbers of differentially expressed lncRNAs (orange) and the proportion to total lncRNAs obtained from reannotation of the gene expression microarrays. B, Pie chart showing the numbers and percentages of differentially expressed lncRNAs in two or more cancer types. C, The top 10 lncRNAs that were significantly upregulated or downregulated in different cancer types.

Figure 3.

A large number of differentially expressed lncRNAs in various cancer types. A, Bar plot showing the numbers of differentially expressed lncRNAs (orange) and the proportion to total lncRNAs obtained from reannotation of the gene expression microarrays. B, Pie chart showing the numbers and percentages of differentially expressed lncRNAs in two or more cancer types. C, The top 10 lncRNAs that were significantly upregulated or downregulated in different cancer types.

Close modal

Secondly, we identified thousands of lncRNAs with potential prognostic significance. Half of the cancer types have more than 2,000 lnRNAs associated with overall survival, including colorectal cancer, breast cancer, lung cancer, and ovarian cancer (Supplementary Fig. S9A). In addition, one third of lncRNAs with prognostic significance are shared by more than one cancer (Supplementary Fig. S9B). The top 10 lncRNAs that were significantly associated with poor prognosis or good prognosis were shown in Supplementary Fig. S9C. Consistent with our study, certain studies have reported that the overexpression of LRRC75A-AS1 is related to the poor prognosis in triple-negative breast cancer (31), whereas the upregulation of C1orf132 is reportedly associated with good prognosis in breast cancer (32). Collectively, lnCAR is a rich resource to discover new cancer biomarkers.

Clinically relevant cancer subtypes revealed by lncRNA expression

We used the lncRNA expression data in the lnCAR database to determine the lncRNA expression subtypes in colorectal, bladder, and breast cancer (Fig. 4A). lncRNA subtype 2 in colorectal cancer resembled the microsatellite instable subtype. lncRNA subtype 1 in bladder cancer mainly resembled the luminal subtype. lncRNA subtype 1 in breast cancer was similar to the basal-like/triple-negative subtype. Although the lncRNA subtypes exhibited a significant correlation with the known subtypes (χ2 test, P < 0.05), the two classification systems offer a distinct stratification of the population (Fig. 4A). Interestingly, the lncRNA subtypes of breast cancer are significantly associated with overall survival (log-rank test, P < 0.05; Fig. 4B), whereas known breast cancer subtypes were not (Supplementary Fig. S10). Moreover, we found the different lncRNA subtypes of breast cancer had distinct profiling of molecular events (Fig. 4C and D). The above results suggest that lncRNAs are useful biomarkers to determine cancer subtypes, further demonstrating the usefulness of lnCAR in cancer research.

Figure 4.

Clinically relevant tumor subtypes revealed by clustering of lncRNA expression. A, The lncRNA expression subtypes exhibit a high concordance with known tumor subtypes in colorectal cancer, bladder cancer, and breast cancer. B, lncRNA expression subtypes appear to be correlated with the relapse-free survival in breast cancer. C, The lncRNA expression patterns of the breast cancer subtypes. D, Enriched KEGG pathways for the protein-coding genes proximal to lncRNA signature.

Figure 4.

Clinically relevant tumor subtypes revealed by clustering of lncRNA expression. A, The lncRNA expression subtypes exhibit a high concordance with known tumor subtypes in colorectal cancer, bladder cancer, and breast cancer. B, lncRNA expression subtypes appear to be correlated with the relapse-free survival in breast cancer. C, The lncRNA expression patterns of the breast cancer subtypes. D, Enriched KEGG pathways for the protein-coding genes proximal to lncRNA signature.

Close modal

lnCAR is a comprehensive open resource for providing the expression profiles and prognostic landscape of lncRNAs derived from reannotation of public microarray data. Compared with other available lncRNA databases, lnCAR has the following advantages (Supplementary Table S8): (i) lnCAR is the first comprehensive database dedicated to reannotating lncRNA probes from expression microarray data, (ii) to the best of our knowledge, lnCAR contains the largest number of cancer samples of any cancer-related lncRNA database, (iii) lnCAR allows users to search any user-defined lncRNAs by genomic coordinates, which will be helpful for exploring the relationship between novel lncRNAs and cancers, and (iv) lnCAR integrates differentially expressed lncRNAs or prognostic lncRNAs in different studies by meta-analysis, promoting the discovery of important lncRNAs in cancers of interest.

In future, as the progressive updating of genomic information, more lncRNA information from the NONCODE database and FANTOM project will be integrated to explore lncRNA expression at the transcript level (33, 34). Besides, we will incorporate more datasets from other cancer types. Furthermore, we will develop an automatic pipeline to check, analyze, and integrate these data, and also encourage users to upload their own data into lnCAR.

No potential conflicts of interest were disclosed.

Conception and design: Y. Zheng, Q. Xu, Z. Zuo, J. Ren

Development of methodology: Y. Zheng, Q. Xu, Z. Zuo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Zheng, H. Hu,

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zheng, M. Liu, H. Hu, Y. Xie

Writing, review, and/or revision of the manuscript: Y. Zheng, Q. Xu, M. Liu, H. Hu, Y. Xie, Z. Zuo, J. Ren

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zheng, Q. Xu, H. Hu, Y. Xie

Study supervision: Q. Xu, H. Hu, Z. Zuo, J. Ren

This work was supported by grants from the National Key R&D Program of China (2017YFA0106700); National Natural Science Foundation of China (91753137, 31471252, 31771462, 81772614, U1611261); Program for Guangdong Introducing Innovative and Entrepreneurial Teams (2017ZT07S096); Guangdong Natural Science Foundation (2014TQ01R387); and Science and Technology Program of Guangzhou, China (201604020003 and 201604046001).

1.
Fatica
A
,
Bozzoni
I
. 
Long non-coding RNAs: new players in cell differentiation and development
.
Nat Rev Genet
2014
;
15
:
7
21
.
2.
Koerner
MV
,
Pauler
FM
,
Huang
R
,
Barlow
DP
. 
The function of non-coding RNAs in genomic imprinting
.
Development
2009
;
136
:
1771
83
.
3.
Rossi
MN
,
Antonangeli
F
. 
LncRNAs: new players in apoptosis control
.
Int J Cell Biol
2014
;
2014
:
473857
.
4.
Shi
X
,
Sun
M
,
Liu
H
,
Yao
Y
,
Song
Y
. 
Long non-coding RNAs: a new frontier in the study of human diseases
.
Cancer Lett
2013
;
339
:
159
66
.
5.
Hajjari
M
,
Salavaty
A
. 
HOTAIR: an oncogenic long non-coding RNA in different cancers
.
Cancer Biol Med
2015
;
12
:
1
9
.
6.
Zhou
Y
,
Zhang
X
,
Klibanski
A
. 
MEG3 noncoding RNA: a tumor suppressor
.
J Mol Endocrinol
2012
;
48
:
R45
53
.
7.
Iyer
MK
,
Niknafs
YS
,
Malik
R
,
Singhal
U
,
Sahu
A
,
Hosono
Y
, et al
The landscape of long noncoding RNAs in the human transcriptome
.
Nat Genet
2015
;
47
:
199
208
.
8.
Li
J
,
Han
L
,
Roebuck
P
,
Diao
L
,
Liu
L
,
Yuan
Y
, et al
TANRIC: an interactive open platform to explore the function of lncRNAs in Cancer
.
Cancer Res
2015
;
75
:
3728
37
.
9.
Ning
S
,
Zhang
J
,
Wang
P
,
Zhi
H
,
Wang
J
,
Liu
Y
, et al
Lnc2Cancer: a manually curated database of experimentally supported lncRNAs associated with various human cancers
.
Nucleic Acids Res
2016
;
44
:
D980
5
.
10.
Chen
G
,
Wang
Z
,
Wang
D
,
Qiu
C
,
Liu
M
,
Chen
X
, et al
LncRNADisease: a database for long-non-coding RNA-associated diseases
.
Nucleic Acids Res
2013
;
41
:
D983
6
.
11.
Zhou
B
,
Zhao
H
,
Yu
J
,
Guo
C
,
Dou
X
,
Song
F
, et al
EVLncRNAs: a manually curated database for long non-coding RNAs validated by low-throughput experiments
.
Nucleic Acids Res
2018
;
46
:
D100
5
.
12.
Cui
T
,
Zhang
L
,
Huang
Y
,
Yi
Y
,
Tan
P
,
Zhao
Y
, et al
MNDR v2.0: an updated resource of ncRNA-disease associations in mammals
.
Nucleic Acids Res
2018
;
46
:
D371
4
.
13.
Liao
Q
,
Liu
C
,
Yuan
X
,
Kang
S
,
Miao
R
,
Xiao
H
, et al
Large-scale prediction of long non-coding RNA functions in a coding-non-coding gene co-expression network
.
Nucleic Acids Res
2011
;
39
:
3864
78
.
14.
Mao
AP
,
Shen
J
,
Zuo
Z
. 
Expression and regulation of long noncoding RNAs in TLR4 signaling in mouse macrophages
.
BMC Genomics
2015
;
16
:
45
.
15.
Barrett
T
,
Wilhite
SE
,
Ledoux
P
,
Evangelista
C
,
Kim
IF
,
Tomashevsky
M
, et al
NCBI GEO: archive for functional genomics data sets–update
.
Nucleic Acids Res
2013
;
41
:
D991
5
.
16.
Gautier
L
,
Cope
L
,
Bolstad
BM
,
Irizarry
RA
. 
affy–analysis of Affymetrix GeneChip data at the probe level
.
Bioinformatics
2004
;
20
:
307
15
.
17.
Lockstone
HE
. 
Exon array data analysis using Affymetrix power tools and R statistical software
.
Brief Bioinform
2011
;
12
:
634
44
.
18.
Ritchie
ME
,
Phipson
B
,
Wu
D
,
Hu
Y
,
Law
CW
,
Shi
W
, et al
limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res
2015
;
43
:
e47
.
19.
Kolde
R
,
Laur
S
,
Adler
P
,
Vilo
J
. 
Robust rank aggregation for gene list integration and meta-analysis
.
Bioinformatics
2012
;
28
:
573
80
.
20.
Benitez-Parejo
N
,
Rodriguez del Aguila
MM
,
Perez-Vicente
S
. 
Survival analysis and Cox regression
.
Allergol Immunopathol
2011
;
39
:
362
73
.
21.
Keck
MK
,
Zuo
Z
,
Khattri
A
,
Stricker
TP
,
Brown
CD
,
Imanguli
M
, et al
Integrative analysis of head and neck cancer identifies two biologically distinct HPV and three non-HPV subtypes
.
Clin Cancer Res
2015
;
21
:
870
81
.
22.
Yu
G
,
Wang
L-G
,
Han
Y
,
He
Q-Y
. 
clusterProfiler: an R package for comparing biological themes among gene clusters
.
OMICS
2012
;
16
:
284
7
.
23.
Gupta
RA
,
Shah
N
,
Wang
KC
,
Kim
J
,
Horlings
HM
,
Wong
DJ
, et al
Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis
.
Nature
2010
;
464
:
1071
6
.
24.
Avazpour
N
,
Hajjari
M
,
Tahmasebi Birgani
M
. 
HOTAIR: a promising long non-coding RNA with potential role in breast invasive carcinoma
.
Front Genet
2017
;
8
:
170
.
25.
Pawłowska
E
,
Szczepanska
J
,
Blasiak
J
. 
The long noncoding RNA HOTAIR in breast cancer: does autophagy play a role?
Int J Mol Sci
2017
;
18
:
pii
:
E2317
.
26.
Zhao
W
,
Luo
J
,
Jiao
S
. 
Comprehensive characterization of cancer subtype associated long non-coding RNAs and their clinical implications
.
Sci Rep
2014
;
4
:
6591
.
27.
Ma
Z
,
Gu
S
,
Song
M
,
Yan
C
,
Hui
B
,
Ji
H
, et al
Long non-coding RNA SNHG17 is an unfavourable prognostic factor and promotes cell proliferation by epigenetically silencing P57 in colorectal cancer
.
Mol Biosyst
2017
;
13
:
2350
61
.
28.
Yang
X
,
Wang
CC
,
Lee
WYW
,
Trovik
J
,
Chung
TKH
,
Kwong
J
. 
Long non-coding RNA HAND2-AS1 inhibits invasion and metastasis in endometrioid endometrial carcinoma through inactivating neuromedin U
.
Cancer Lett
2018
;
413
:
23
34
.
29.
Zhang
J
,
Fan
D
,
Jian
Z
,
Chen
GG
,
Lai
PB
. 
Cancer specific long noncoding RNAs show differential expression patterns and competing endogenous RNA potential in hepatocellular carcinoma
.
PLoS One
2015
;
10
:
e0141042
.
30.
Yang
Y
,
Chen
L
,
Gu
J
,
Zhang
H
,
Yuan
J
,
Lian
Q
, et al
Recurrently deregulated lncRNAs in hepatocellular carcinoma
.
Nat Commun
2017
;
8
:
14421
.
31.
Lv
M
,
Xu
P
,
Wu
Y
,
Huang
L
,
Li
W
,
Lv
S
, et al
LncRNAs as new biomarkers to differentiate triple negative breast cancer from non-triple negative breast cancer
.
Oncotarget
2016
;
7
:
13047
59
.
32.
Wu
L
,
Xu
Q
,
Zhang
H
,
Li
M
,
Zhu
C
,
Jiang
M
, et al
A new avenue for obtaining insight into the functional characteristics of long noncoding RNAs associated with estrogen receptor signaling
.
Sci Rep
2016
;
6
:
31716
.
33.
Hon
CC
,
Ramilowski
JA
,
Harshbarger
J
,
Bertin
N
,
Rackham
OJ
,
Gough
J
, et al
An atlas of human long non-coding RNAs with accurate 5′ ends
.
Nature
2017
;
543
:
199
204
.
34.
Fang
S
,
Zhang
L
,
Guo
J
,
Niu
Y
,
Wu
Y
,
Li
H
, et al
NONCODEV5: a comprehensive annotation database for long non-coding RNAs
.
Nucleic Acids Res
2018
;
46
:
D308
14
.

Supplementary data