To obtain high-confidence lncRNAs, the following steps were adopted after the assemblied transcripts were merged from over 2000 RNA-Seq libraries. The merged transcripts were annotated using the Cuffcompare program to filter out the known protein-coding transcripts, rRNA and tRNA with the comparison code '='; Questionable transcripts were detected using Cuffcompare withthe comparison codes ‘e’, ‘p’ and ‘s’;Transcripts with lengths more than 200nt were selected as lncRNA candidates; Transcripts with FPKM scores smaller than 0.5 in all samples were discarded; Further screened through the protein-coding score test using Coding Potential Calculator (CPC2), PlncPRO and Protein family database (Pfam). Transcripts identified as lncRNAs by all the three algorithms were retained.

       The prediction of assoications of lncRNAs with agricultural traits was based on the overlap of lncRNAs with GWAS tag SNPs and QTLs. GWAS information of rice were from Rice SNP-Seek Database and QTLs information were downloaded from Q-TARO (QTL Annotation Rice Online) database (Yamamoto et al. 2012). The GWAS SNP contributed phenotype was predicted as possible function of a specific lncRNA if that GWAS tag SNPs co-located with this lncRNA. Accordingly, when a specific lncRNA locus overlapped with a QTL, this QTL-related trait was tagged on this lncRNA.

       To profile the expression level of lncRNAs, the expression value in all the 2313 RNA-Seq library samples were calculated, which can be downloaded in the download page. From the overall expression profile, over 10 expression datasets were selected for visualization of lncRNAs expression and prediction of functions. These expression profile covers different tissues in indica and japonica, different growth conditions such as including Phosphate starvation, salt stress, cadmium stress, drought stress, cold stress, osmotic stress and flood stress as well as samples grown under different hormone treatments, covering JA treatment and ABA treatment. Then in each expression datasets, τ-value and cv (coefficient of variance) value were used to determine HK (housekeeping) lncRNAsand (τ-value <= 0.5 and cv <= 0.5) and TS (tissue-specific), ), SR (stress responsive) lncRNAs (τ-value >= 0.95).

       For each lncRNA in RiceLncPedia, the basic molecular features are provided including ID, sequences, genome position, GC content, coding potential and classification. Additionally, the following informations can be retrieved, containing expression profile, genome variation, associations with phenotype based on GWAS and QTLs, overlap with transposons, lncRNA-miRNA interactions and miRNA precursors.

       In the download page, all the lncRNAs sequences in fasta format, gtf file and relevant multi-omics annotation can be downloaded.