how to combine junction.bed files with different rows into one table

[shirley0818]

Hi All,

I am working on ~100 samples for detecting alternative splicing. TopHat generates a junction.bed file for each sample. However, each of these bed files has different number of rows, and the coordinates of each junction is not same across samples. I think this junction.bed file includes known and novel junctions.

Since I am only interested in known junctions in Ensembl annotation database, how can I map these 100 junction.bed files to Ensembl gtf file and obtain a table matrix with the row as known-junction and column as sampleID?

Or do I need to create a exon-exon junction annotation bed file from Ensembl, then apply RSeQC to obtain reads for each junction against mapped .bam files?

Many thanks,

Shirley

Shirley

[GenoMax]

If I understand your aim correctly between the following two threads you should be able to get pointers for what you need.

gtftobed: https://www.biostars.org/p/56280/
bedops: https://www.biostars.org/p/119835/

[shirley0818]

Thanks GenoMax for your quick response. I have tried "bedtools" as you suggested, but got very strange results: Below are the example files and output.

1. Here is s1.junction.bed file generated by TopHat
Column #5 "score" is the number of reads that contain the junction.
1 12197 12639 JUNC00000001 1 + 12197 12639 255,0,0 2 30,45 0,397
1 12190 12686 JUNC00000002 7 + 12190 12686 255,0,0 2 37,74 0,422
1 12633 13292 JUNC00000003 6 + 12633 13292 255,0,0 2 64,72 0,587

2. Here is hg19_RefSeq.mod.sorted.bed
1 11873 12047 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12210 12684 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12644 13464 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12627 13527 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12661 13465 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12663 13469 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12657 13539 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,

3. run bedtools
bedtools intersect -a hg19_RefSeq.sorted.bed -b s1.junctions.bed > out.txt

The results in out.txt is very strange since column #5 "score" (read counts) is all 0.

1 11873 12047 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12210 12684 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12644 13464 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,
1 12627 13527 NR_046018 0 + 14409 14409 0 3 354,109,1189, 0,739,1347,

Could you let me know whether I have misused "bedtools intersect"?

Many thanks,

[shirley0818]

Hi GenoMax,

I have figured it out by adding -wa -wb option:
bedtools intersect -wa -wb -a A.bed -b B.bed

Many thanks,
Shirley