Assembly of transcripts from junction reads

Question

[markddesimone]

I have a series of files containing exon-exon junctions, essentially chr, start, end, strand of each intron. Is there a way to assemble these into transcripts without having the original bam files?

Thank you

Answer 1

[zubenel]

I think it would be easier for someone to help if you could provide us with an example of some lines of your files. And could you describe how these files were generated?

[markddesimone]

Thank you, here is an example, they are bam.bed files, no header, the usual bed format. After adjusting start end as described here Tophat Junctions.Bed File the coordinates are the intron coordinates. Only junctions between exons are detected. I can use gencode annotation to determine which exons are likely either side of the intron but would like to assemble them into likely transcripts so I could get transcript abundance measures. :

1   12147   12630   JUNC00000001    1   +   12147   12630   255,0,0 2   80,18   0,465
1   12680   13510   JUNC00000002    1   +   12680   13510   255,0,0 2   41,58   0,772
1   14730   15062   JUNC00000005    92  -   14730   15062   255,0,0 2   99,93   0,239
1   14939   15891   JUNC00000004    93  -   14939   15891   255,0,0 2   99,96   0,856
1   15903   16703   JUNC00000007    27  -   15903   16703   255,0,0 2   44,97   0,703
1   16292   16688   JUNC00000008    1   -   16292   16688   255,0,0 2   18,82   0,314
1   16967   17323   JUNC00000009    21  -   16967   17323   255,0,0 2   88,91   0,265
1   17272   17702   JUNC00000010    30  -   17272   17702   255,0,0 2   96,97   0,333
1   17647   18011   JUNC00000011    52  -   17647   18011   255,0,0 2   95,97   0,267
1   17971   18365   JUNC00000012    3   -   17971   18365   255,0,0 2   90,98   0,296
1   17999   24813   JUNC00000013    4   -   17999   24813   255,0,0 2   62,76   0,6738
1   18273   24826   JUNC00000014    10  -   18273   24826   255,0,0 2   93,89   0,6464

[zubenel]

Maybe RegTools could help: https://regtools.readthedocs.io/en/latest/commands/junctions-annotate/ I haven't tried this command but it seems that it can output transcripts and genes that overlap junctions if you have provided reference fasta and annotations gtf files.