Hello all,
I am a total noob in analyzing NGS data, and I would really appreciate if I can get help with my prob. Also, kindly excuse me if I'm using the wrong terminology to describe something.
My experiment:
Analysis of differences in DNA methylation across genome. Methylated DNA was pulled down using 5-methyl C antibody, and sequencing was performed using SOLiD. 4 samples were processed, as follows:
1. Control, Initial Timepoint
2. Test, Initial Timepoint
3. Control, Final Timepoint
4. Test, Final Timepoint
All raw data was analyzed and filtered by the people who helped me operate the machine, and they finally gave me .bam files, which I opened using IGV genome browser.
My problem:
The fold coverage for the third sample is relatively low, while all the others are good. Because of this, when I see regions where the number of hits are different between samples, I don't know if it's because of the differences in fold coverage or they are genuinely differently methylated. In other words, I have no control over the number of false positives I'm detecting. This was more apparent when I tried to manually validate the results by bisulfite sequencing, as I couldn't see differences in methylation even at regions that showed largest difference in SOLiD data.
My question:
Firstly, what is the way to normalize the coverage for all samples? Is there a way I can find out a multiplication factor which I can use to multiply the hits I get in sample 3 to normalize the coverage? In addition, what is the flowchart or pipeline for me to analyze this data automatically - normalization of coverage, detection of false positives, and generation of a list of genuine targets. I can say I have a bit of bioinformatics and programming knowledge so I'm fairly open to using scripts or any other similar tactics.
I'm sorry for rambling on so much. I will appreciate any kind of help that I can get with this problem.
Thank you all,
TEJ
I am a total noob in analyzing NGS data, and I would really appreciate if I can get help with my prob. Also, kindly excuse me if I'm using the wrong terminology to describe something.
My experiment:
Analysis of differences in DNA methylation across genome. Methylated DNA was pulled down using 5-methyl C antibody, and sequencing was performed using SOLiD. 4 samples were processed, as follows:
1. Control, Initial Timepoint
2. Test, Initial Timepoint
3. Control, Final Timepoint
4. Test, Final Timepoint
All raw data was analyzed and filtered by the people who helped me operate the machine, and they finally gave me .bam files, which I opened using IGV genome browser.
My problem:
The fold coverage for the third sample is relatively low, while all the others are good. Because of this, when I see regions where the number of hits are different between samples, I don't know if it's because of the differences in fold coverage or they are genuinely differently methylated. In other words, I have no control over the number of false positives I'm detecting. This was more apparent when I tried to manually validate the results by bisulfite sequencing, as I couldn't see differences in methylation even at regions that showed largest difference in SOLiD data.
My question:
Firstly, what is the way to normalize the coverage for all samples? Is there a way I can find out a multiplication factor which I can use to multiply the hits I get in sample 3 to normalize the coverage? In addition, what is the flowchart or pipeline for me to analyze this data automatically - normalization of coverage, detection of false positives, and generation of a list of genuine targets. I can say I have a bit of bioinformatics and programming knowledge so I'm fairly open to using scripts or any other similar tactics.
I'm sorry for rambling on so much. I will appreciate any kind of help that I can get with this problem.
Thank you all,
TEJ