gff3-bed-bam for RNAseq

[Hagen]

Dear all,

I have some splice junctions (in GFF3) that I ultimately want to convert to bam. I was thinking to convert first to bed myself and then to bam using bedToBam.
To test the second step I grabbed a bed-example from teh UCSC browser (each line encodes TWO exons) and ran a little test on it, converting to bam and then back to bed. This results in information loss:

nuvol:cufflinks htilgner$ cat test.bed
chr22 1000 5000 cloneA 960 + 1000 5000 0 2 567,488, 0,3512
chr22 2000 6000 cloneB 900 - 2000 6000 0 2 433,399, 0,3601
nuvol:cufflinks htilgner$ bedToBam -i test.bed -g hg19.genome.file > test.bam
nuvol:cufflinks htilgner$ bamToBed -i test.bam
chr22 1000 5000 cloneA 255 +
chr22 2000 6000 cloneB 255 -

Do you know why ? More importantly, is there an appropriate way to encode a spliced read in bam format ? How would that work ? Alternatively, is there a way to convert gff3 to bam directly ?

Thanks

[dpryan]

Given that the bedToBam program is general purpose, I'd be pretty surprised if it knew enough about what the various meta-information columns meant in order to store them in a meaningful way in BAM format. Spliced read are generally denoted with Ns in the CIGAR string.

[Hagen]

Dear dpryan,

thanks for your answer. Do you have a way to force the below 2 example read (that are spliced) into a bam file ?
chr22 1000 5000 cloneA 960 + 1000 5000 0 2 567,488, 0,3512
chr22 2000 6000 cloneB 900 - 2000 6000 0 2 433,399, 0,3601

Best & thanks

[dpryan]

Not off-hand, I've never needed such a file. It'd be easy enough to just write a conversion program in C that would parse the BED file for the exonic regions, grab the relevant sequence from a fasta file, and write everything to a BAM file. I haven't a clue why you would want such a file, but that would work.

[Hagen]

OK, I get it now, I believe. My problem was that I hoped not to get into teh specific of sam/bam and simply convert ... will write the converter ...
thanks for your help!