Tweet
Hi all! I am very new to this field, so I apologize in advance for my ignorance. I have done several chip-seq experiments on a protein from A.thaliana that functions in homologous recombination. I am struggling a lot with analyzing my data, mostly because I do not completely understand how the tools work. I use galaxy to do the analysis, Bowtie and MACS.
I have several questions regarding these tools, hopefully someone will help!
1. I know MACS provides the peak scores as -10log(pvalue) and it gives me ranges usually from 50-3100. But how does MACS2 calculate the score? Using the same dataset, I get very narrow ranges usually in single digits.
2. What do treatment wig and control wig files show? I read that they show the binding signal, however this explanation is way to vague for me. And how can I correlate the peak bed file to the bigwig file on a genome browser?
3. When I use MACS to call the peaks, I use default setting (pvalue cutoff 1e-5, and MFOLD value of 32). However, MACS fails everytime since it cannot find enough paired peaks to build the shifting model. Lowering the MFOLD value does not help. Only when I choose not to build the shifting model, only then will MACS succeed in calling the peaks. But, now I am worried, can I trust these peaks that MACS called? If it cannot make a shifting model, how could I predict the location of the binding site? I leave the arbitrary shift size 100 bp, but how do I know if this is the proper value to use? I also read that MACS cannot build the shifting model since there is no defined peaks, this can be a result of broad peaks (like for RNApol), it looks like my peaks are broad as well.
Thank you for your help!!!!
I have several questions regarding these tools, hopefully someone will help!
1. I know MACS provides the peak scores as -10log(pvalue) and it gives me ranges usually from 50-3100. But how does MACS2 calculate the score? Using the same dataset, I get very narrow ranges usually in single digits.
2. What do treatment wig and control wig files show? I read that they show the binding signal, however this explanation is way to vague for me. And how can I correlate the peak bed file to the bigwig file on a genome browser?
3. When I use MACS to call the peaks, I use default setting (pvalue cutoff 1e-5, and MFOLD value of 32). However, MACS fails everytime since it cannot find enough paired peaks to build the shifting model. Lowering the MFOLD value does not help. Only when I choose not to build the shifting model, only then will MACS succeed in calling the peaks. But, now I am worried, can I trust these peaks that MACS called? If it cannot make a shifting model, how could I predict the location of the binding site? I leave the arbitrary shift size 100 bp, but how do I know if this is the proper value to use? I also read that MACS cannot build the shifting model since there is no defined peaks, this can be a result of broad peaks (like for RNApol), it looks like my peaks are broad as well.
Thank you for your help!!!!