[BioC] CRISPRseek question on scoring
Zhu, Lihua (Julie)
Julie.Zhu at umassmed.edu
Wed Aug 13 17:26:35 CEST 2014
The score from the website would be,
100/( 100 + [CRISPRseek top100OfftargetTotalScore ]) = 100/(100+ 10) = 90.9
On 8/13/14 11:09 AM, "Andy Crouse" <andy at transomic.com> wrote:
Thank you Julie!
So, if I understand correctly, then if there were to be a top100offtargetTotalScore column in the CRISPRseek summary file and the score there was 10, then the crispr.mit.edu <http://crispr.mit.edu> site would return a score of 90?
And I can expect that (100 [CRISPRseek top5OfftargetTotalScore ]) to be greater than the score from the MIT site?
Thank you again! This is very helpful and I really like the flexibility of the module.
From: Zhu, Lihua (Julie) [mailto:Julie.Zhu at umassmed.edu]
Sent: Wednesday, August 13, 2014 9:16 AM
To: Andy Crouse
Cc: bioconductor at r-project.org
Subject: Re: CRISPRseek question on scoring
Thanks for the feedback!
I assume you are comparing the scores in the summary file not the individual offtarget output file. The score for each individual offtarget should be the same between these two. However the summary score of CRISPRseek has difference meaning from that of the web application. As a summary, CRISPRseek adds the topN per users choice of topN.OfftargetTotalScore, eg., 5 or 10 off target scores together, reasoning that the top off-targets are the most critical ones. The CRISPR design website ranks guides based on an aggregate score using the top 100 matches (details at http://crispr.mit.edu/about - under "aggregate scores"). It is important to note that their scale goes in the opposite direction, where scores start at 100% and subtract the weighted off target score, whereas our summary score starts at zero and adds the individual score of each off target.
Hope that is clear. Please let me know if you need additional information.
On 8/13/14 1:57 AM, "Andy Crouse" <andy at transomic.com> wrote:
I am working with your CRISPRseek module. I plugged in the example sequence to the CRISPR.MIT.EDU <http://CRISPR.MIT.EDU> <http://CRISPR.MIT.EDU> site and it selects the same gRNA, but the scores are very different.
The CRISPRseek guide says that the default uses the same weighting as Hsu et al. The website scores are between 48 and 72 while the CRISPRseek scores are between 5 and 20. Am I interpreting something incorrectly or do I have setting wrong?
This is what I ran:
offTargetAnalysis(inputFilePath, findgRNAsWithREcutOnly = FALSE,
REpatternFile = REpatternFile,findPairedgRNAOnly = FALSE,
BSgenomeName = Hsapiens,
txdb = TxDb.Hsapiens.UCSC.hg19.knownGene,
max.mismatch = 4, outputDir = outputDir, overwrite = TRUE)
Thanks for any assistance here.
Andy Crouse, Ph.D.
Senior Product Manager
[[alternative HTML version deleted]]
More information about the Bioconductor