• Home
• About

About TENOR

"TENOR (Transcriptome ENcyclopedia Of Rice)" is a database to privide transcriptional activity on the rice genome at the nucleotide level based on the RNA-Seq data under 140 environmental stresses and plant hormone treated conditions. As well as expression profiles, information of cis-regulatory elements in promoter regions and co-expressed transcript are provided for each transcript.

About samples and mRNA-Seq data

Treatment conditions

The rice (Oryza sativa L. ssp. japonica cv. Nipponbare) seedlings were used for all experiments. All samples were hydroponically grown at 28ºC temperatures, 16h-light/8h-dark cycle with the light period from 6:00 A.M. to 10:00 P.M., and 70-80% relative humidity. 0h (control) samples were harvested at 9:00 A.M.

• Mizuno et al. 2010 [pubmed] [DRA]
• Oono et al. 2011 [journal] [DRA]
• Oono et al. 2014 [pubmed] [DRA]
• 10 days after high/low phosphate stress condition followed by 1 day under normal condition for recoverly.

mRNA-Seq data

Database Contents and Functions

Entrances to TENOR

The following three entrances are available to access expression data in TENOR.

• Search transcripts by kewords (e.g. transcript ID, definition)
• Go to GBrowse where transcript structures with expression profiles are displayed along the rice genome. Transcriptional activity with nucleotide resolution and cis-regulatory elements in the upstream regions of each transcript are also provided.
• Search transcripts by their responsive expression pattern ("treatment type", "tissue", "timepoint", "direction (up/down) of responsive pattern").

(A) Keyword search

This function will return the list of transcript IDs that contain "search words" in their ID or description fields. Responsive expression patterns (fold-change and FDR) for each condition are displayed as well as transcript IDs and descriptions. Each transcript has four links to the following external information.

• GBrowse (Genome Browser): Gene structures, transcriptional activity (depth of coverage of RNA-Seq reads) and other genomic information (PLACE cis-regulatory elements, repetitive regions, etc.) are shown along chromosome.
• Expression profile viewer: It provides expression profiles under the different environmental conditions.
• RAP-DB: Rice Annotation Project Database provides genome annotations and other functional information for each RAP-DB transcript.
• Co-expression viewer: It provides a list of co-expressed transcripts with a selected reference transcript based on both Pearson's and Gini correlation coefficient values.

(B) Genome Browser

(C) Responsive Gene Search

Users can search transcripts by their responsive expression patterns and thresholds. A list of transcripts whose expression pattern is consistent with specified responsive expression pattern will be shown in the bottom frame as in the keyword search.

Gene prediction program PARPNTE

PARPNTE can predict both protein-coding and non-coding transcript structures simultaneously with coding sequence (CDS) information based on the alignment of mRNA-Seq reads on the genome and the hidden Markov model (HMM) trained with RAP-DB representative transcripts with FLcDNA evidence.

Performance comparison with Cufflinks2

Sensitivity (Sn) and Specificity (Sp) were compared betwee PARPNTE and Cufflinks2 using cuffcompare.
mRNA-Seq data in TENOR and representative transcripts in RAP-DB were used for the gene prediction by two programs.

	PARPNTE:
	#     Query mRNAs :   48366 in   38228 loci  (30293 multi-exon transcripts)
	#            (5183 multi-transcript loci, ~1.3 transcripts per locus)
	# Reference mRNAs :   43466 in   37870 loci  (30911 multi-exon)
	
	#--------------------|   Sn   |  Sp   |  fSn |  fSp
	        Base level:      51.2    64.1     -       -
	        Exon level:      52.4    58.9    54.6    61.3
	      Intron level:      70.1    80.6    71.8    82.6
	Intron chain level:      32.5    33.1    57.3    58.5
	  Transcript level:       0.0     0.0     0.1     0.1
	       Locus level:      26.2    25.9    34.1    33.2
	
	     Matching intron chains:   10035
	              Matching loci:    9907
	
	          Missed exons:   47549/173117  ( 27.5%)
	           Novel exons:   25775/154037  ( 16.7%)
	        Missed introns:   33323/125962  ( 26.5%)
	         Novel introns:   14557/109525  ( 13.3%)
	           Missed loci:   15352/37870   ( 40.5%)
	            Novel loci:   12423/38228   ( 32.5%)
	Cufflinks:
	#     Query mRNAs :   45415 in   30617 loci  (30249 multi-exon transcripts)
	#            (9625 multi-transcript loci, ~1.5 transcripts per locus)
	# Reference mRNAs :   43466 in   37870 loci  (30911 multi-exon)
	
	#--------------------|   Sn   |  Sp   |  fSn |  fSp
	        Base level:      59.0    50.8     -       -
	        Exon level:      49.1    54.2    51.4    56.7
	      Intron level:      69.3    79.5    71.4    81.9
	Intron chain level:      31.3    32.0    55.3    56.5
	  Transcript level:       0.0     0.0     0.0     0.0
	       Locus level:      25.1    31.0    34.4    40.6
	
	     Matching intron chains:    9681
	              Matching loci:    9513
	
	          Missed exons:   44641/173117  ( 25.8%)
	           Novel exons:   20563/156764  ( 13.1%)
	        Missed introns:   34431/125962  ( 27.3%)
	         Novel introns:   15408/109802  ( 14.0%)
	           Missed loci:   13968/37870   ( 36.9%)
	            Novel loci:    5892/30617   ( 19.2%)
		

Download PARPNTE

Please download from "Data Downloads" page

How to use PARPNTE

PARPNTE predicts gene structures and protein-coding regions based on mapping data of mRNA-Seq reads and genome sequence data.
The prediction depends on parameters generated from genome sequnece, splicing junction and depth of mRNA-Seq reads and reference gene annotation data.

Step 1) Building HMM

1. Prepare input files for training

- Reference genome sequence (FASTA)
- splicing junction data (BED format file reported by TopHat)
- read depth file (output file of SAMtools "depth" command with accepted_hits.bam file reported by TopHat)
- reference annotation (GFF)

2. Run building HMM

$ java -jar parpnte.jar -ms -g -q -d -j -sb

- socrefiles_basename_path : Set the name of the directory and prefix in which PARPNTE will write all of its output.
If you set /home/bio/parpnte/rice, all output files (e.g., rice.depth_score, rice.base_score, rice.trans_score) will be generated under the "/home/bio/parpnte" direcotry.

3. Output files

- transcripts.gff : result of gene prediction
- amino_acid_sequences.fasta : amino acid sequneces of predicted genes
- cds_sequences.fasta : nucleotide sequences of protein coding regions
- mrna_sequences.fasta : nucleotide sequences of transcribed regions
- junctions_modified.bed : splicing junction data (BED) after removing duplicated and noise data
- junctions_removed.bed : splicing junction data (BED) removed
- run.log : running log file for gene prediction
- error.log : error log file for gene prediction

Step 2) Gene prediction

1. Prepare input files for gene prediction

- Reference genome sequence (FASTA)
- splicing junction data (BED format file reported by TopHat)
- read depth file (output file of SAMtools "depth" command with accepted_hits.bam file reported by TopHat)
- basename for scoring matrix generated in the Building HMM step (e.g., /home/bio/parpnte/rice)

2. Run gene prediction

$ java -jar parpnte.jar -g -d -j -sb -o -chr

3. Output files

- transcripts.gff : result of gene prediction
- amino_acid_sequences.fasta : amino acid sequneces of predicted genes
- cds_sequences.fasta : nucleotide sequences of protein coding regions
- mrna_sequences.fasta : nucleotide sequences of transcribed regions
- junctions_modified.bed : splicing junction data (BED) after removing duplicated and noise data
- junctions_removed.bed : splicing junction data (BED) removed
- run.log : running log file for gene prediction
- error.log : error log file for gene prediction

Licence

This software is released under the MIT License.