problem in variant calling with rna-seq

[bongbimit]

hi all,
i have a question about my problem in rna-seq with my sample.
i have 25 different sample from 25 human and i sequence them by rna-seq to find variant calling by Tophat,mpileup and VarScan.
My result is also good when some sample's SNP are found on dbSNP, but i have a problem when some sample have SNP in intron region(many reads is mapping into this SNP, it 's not a noise), how can i explain this problem ? is it noise in sequencing or is there some mechanism intron region can come with mature mRNA in human ?

Thank you all.

[Brian Bushnell]

The annotations are not completely correct, so perhaps the exon/intron boundary is incorrect, or there is a previously unknown gene or exon there. Furthermore, some mutations can break splice sites so that they are missed. RNA-seq data often has a few reads that will map to non-exonic areas, possibly for other reasons... but usually not many to the same location.

[bongbimit]

The annotations are not completely correct, so perhaps the exon/intron boundary is incorrect, or there is a previously unknown gene or exon there. Furthermore, some mutations can break splice sites so that they are missed. RNA-seq data often has a few reads that will map to non-exonic areas, possibly for other reasons... but usually not many to the same location.

i think annotation which i had is correct (hg19 homo sapiens) and the % mapping of this snp is also high, but i will check your opinion about mutation because my mRNA is oncogene (cancer)

[Brian Bushnell]

i think annotation which i had is correct (hg19 homo sapiens) and the % mapping of this snp is also high, but i will check your opinion about mutation because my mRNA is oncogene (cancer)

I don't mean the annotation is "wrong", I just mean that it's neither complete nor perfect, just like the HG19 genome. And your note that the source is cancer does make the mutation theory more plausible.