Hi everyone
I've been running some transcription factor binding site analysis using the Storm and Motifclass utilities in the CREAD package. I got some interesting results using a functional depth cut-off and decided to run my data through again this time using a p-value cut-off as a sort of sanity check. As always with this sort of analysis, this opened a huge can of worms because there is very little agreement between results. I've got a couple questions about this that I was hoping some of you could answer:
1. I know that altering parameters in one's analysis can lead to different results. Should I be surprised/concerned that my results were so different?
2. Has anyone used Storm/Motifclass? What did you find were the best parameters for your particular data-set?
3. I've also played around with DME/Matcompare and Patser. Are there any PWM or non-PWM based transcription factor binding site analysis methods that you really like working with?
I've been running some transcription factor binding site analysis using the Storm and Motifclass utilities in the CREAD package. I got some interesting results using a functional depth cut-off and decided to run my data through again this time using a p-value cut-off as a sort of sanity check. As always with this sort of analysis, this opened a huge can of worms because there is very little agreement between results. I've got a couple questions about this that I was hoping some of you could answer:
1. I know that altering parameters in one's analysis can lead to different results. Should I be surprised/concerned that my results were so different?
2. Has anyone used Storm/Motifclass? What did you find were the best parameters for your particular data-set?
3. I've also played around with DME/Matcompare and Patser. Are there any PWM or non-PWM based transcription factor binding site analysis methods that you really like working with?
Comment