I have two cell lines, one of which has a transfected, non-reference antibiotic resistance gene. In one adjacent gene (which exist in both cell lines), there are two possible heterozygous SNPs. The possible alleles of the resistance gene / SNP gene are (without going into details as to why) as follows:
cell line 1:
cell line 2:
The original alignment I made (Tophat) used the normal hg19 reference, which does not contain the resistance gene, and I can thus not see any reads that map to that sequence. What I need to be able to do is to count the number of reads that map to the different alternatives outlined above, as part of some troubleshooting work, and I am unsure how I would do this.
One of the problems is that my reads are only 100 bp long, while the resistance gene is approximately 80 bp away from the farthest situated SNP, which means that I'm having a hard time counting reads covering those specific sequences - I've used samtools and grep for this, with little success.
I have used BBduk with a custom reference sequence (only covering the resistance gene and the gene of interest) followed by alignment with BBmap and viewing in IGV. I can here clearly see if reads are present in the resistance genes for the different cell lines, which behaves as I expect, but the alignment / reads seem "less precise" than the ones I get from the normal alignment with Tophat, i.e. there are a lot of seemingly random mutations in the reads that still map to the reference. I have tried playing with the number of allowed mismatches to circumvent this, but allowing 1 mismatch does not give significantly different results from using 2, and using 0 gives extremely few reads to work with.
How would you go about getting this kind of specific read counts for data containing non-reference sequences? Do I need to re-align the data with a modified reference sequence containing the antibiotic resistance gene? If so, should I align both my cell lines to that reference, or just the one that contains it? What are any other possible alternatives?
cell line 1:
- resistance, SNP 1
- resistance, SNP 2
- no resistance, SNP 2
- no resistance, no SNPs
cell line 2:
- no resistance, SNP 2
- no resistance, no SNPs
The original alignment I made (Tophat) used the normal hg19 reference, which does not contain the resistance gene, and I can thus not see any reads that map to that sequence. What I need to be able to do is to count the number of reads that map to the different alternatives outlined above, as part of some troubleshooting work, and I am unsure how I would do this.
One of the problems is that my reads are only 100 bp long, while the resistance gene is approximately 80 bp away from the farthest situated SNP, which means that I'm having a hard time counting reads covering those specific sequences - I've used samtools and grep for this, with little success.
I have used BBduk with a custom reference sequence (only covering the resistance gene and the gene of interest) followed by alignment with BBmap and viewing in IGV. I can here clearly see if reads are present in the resistance genes for the different cell lines, which behaves as I expect, but the alignment / reads seem "less precise" than the ones I get from the normal alignment with Tophat, i.e. there are a lot of seemingly random mutations in the reads that still map to the reference. I have tried playing with the number of allowed mismatches to circumvent this, but allowing 1 mismatch does not give significantly different results from using 2, and using 0 gives extremely few reads to work with.
How would you go about getting this kind of specific read counts for data containing non-reference sequences? Do I need to re-align the data with a modified reference sequence containing the antibiotic resistance gene? If so, should I align both my cell lines to that reference, or just the one that contains it? What are any other possible alternatives?