analyzing TIDE results
2,054 views
Skip to first unread message
regulatingge...@gmail.com
Jul 27, 2016, 1:37:57 PM7/27/16
to Genome Engineering using CRISPR/Cas Systems
Hi,
I came across the TIDE online tool for analyzing indel events. Does anyone know how to interpret the output? and how accurate is it in comparison to Surveyor?
Thanks.
I came across the TIDE online tool for analyzing indel events. Does anyone know how to interpret the output? and how accurate is it in comparison to Surveyor?
Thanks.
Mike
Jul 28, 2016, 10:34:03 AM7/28/16
to Genome Engineering using CRISPR/Cas Systems
Hi,
I've found TIDE very useful. I think it is much more useful than Surveyor. This is because Surveyor relies on mismatches in annealed PCR products, so if you have a for example, a clone with different indels in each allele, it will look the same as a clone with one indel allele and one wildtype allele in a surveyor/T7EI assay. TIDE can discriminate between these two possibilities, as well as giving you information about the type of indel present. I've completely stopped T7EI since discovering TIDE.
As for helping you interpret the results, try loading the sample data on the webtool. If you still have trouble, upload a screenshot of your results and describe which parts of the output you are struggling with and I'll try to help you.
Regards,
Mike
I've found TIDE very useful. I think it is much more useful than Surveyor. This is because Surveyor relies on mismatches in annealed PCR products, so if you have a for example, a clone with different indels in each allele, it will look the same as a clone with one indel allele and one wildtype allele in a surveyor/T7EI assay. TIDE can discriminate between these two possibilities, as well as giving you information about the type of indel present. I've completely stopped T7EI since discovering TIDE.
As for helping you interpret the results, try loading the sample data on the webtool. If you still have trouble, upload a screenshot of your results and describe which parts of the output you are struggling with and I'll try to help you.
Regards,
Mike
Message has been deleted
jean-pierre de villartay
Aug 2, 2016, 5:37:08 AM8/2/16
to Genome Engineering using CRISPR/Cas Systems
TIDE is a very powerfull tool
If you work with clones, you can very nicely see your 2 alleles. If you only see one (a single bar at about 50%) then consider increasing the size of detectable indels (set up to 10 by default) and you'll be surprised.......
If you work with bulk populations, the sensitivity tends to drop as the complexity of your population increases. For exemple for a gRNA that give me about 100% indels as determined by the loss of a restriction site sitting on the target, TIDE calculated 70% mutagenesis only
Hope this helps
If you work with clones, you can very nicely see your 2 alleles. If you only see one (a single bar at about 50%) then consider increasing the size of detectable indels (set up to 10 by default) and you'll be surprised.......
If you work with bulk populations, the sensitivity tends to drop as the complexity of your population increases. For exemple for a gRNA that give me about 100% indels as determined by the loss of a restriction site sitting on the target, TIDE calculated 70% mutagenesis only
Hope this helps
jane.elea...@gmail.com
Aug 4, 2016, 10:52:48 AM8/4/16
to Genome Engineering using CRISPR/Cas Systems
I am also a huge fan of TIDE. If I TOPO clone and sequence the alleles from single clones, I get exactly the indels detected by TIDE. In my transient tranfections I can get around 75% efficiency. It really has made life a whole lot easier!
regulatingge...@gmail.com
Aug 4, 2016, 5:24:22 PM8/4/16
to Genome Engineering using CRISPR/Cas Systems
Hi,
do you mean 75% efficiency from total transfected cell population? When I ran TIDE on cells 24 hours after transfection, the best gRNA (of the 3 tested) had 8% efficiency. Do you think this means it is worthless to go through picking colonies?
Thanks
do you mean 75% efficiency from total transfected cell population? When I ran TIDE on cells 24 hours after transfection, the best gRNA (of the 3 tested) had 8% efficiency. Do you think this means it is worthless to go through picking colonies?
Thanks
jane.elea...@gmail.com
Aug 5, 2016, 7:46:11 AM8/5/16
to Genome Engineering using CRISPR/Cas Systems
Hi,
I make genomic DNA 40-48h after transfection. I'm using the EGFP vector and the signal (and hence Cas9 expression) is better on the second day. Is your tranfection efficiency OK? I wouldn't personally use a gRNA at 8% efficiency but I've used it at 40% and that was OK.
I make genomic DNA 40-48h after transfection. I'm using the EGFP vector and the signal (and hence Cas9 expression) is better on the second day. Is your tranfection efficiency OK? I wouldn't personally use a gRNA at 8% efficiency but I've used it at 40% and that was OK.
Fatwa Adikusuma
Aug 6, 2016, 4:35:52 AM8/6/16
to Genome Engineering using CRISPR/Cas Systems
I use RFLP and TIDE a lot. Most of the time the RFLP shows no indication of WT present which means efficiency should be close to 100%. The TIDE analysis also shows 0% of WT (0). However, the total efficiency calculated by TIDE is pretty low (let's say 60%), and the R-squared is also low (0.60). When the R-squared is high (e.g. 0.90), the TIDE-calculated total efficiency is high as well (90% with 0% of WT). I don't know a lot about statistic, but what I think about the real efficiency should be calculated this way --> TIDE-calculated efficiency/R-squared x 100%. For instance for the first case --> 60%/0.6 x 100% = 100%.
I hope people who created TIDE can give more explanation here.
Cheers,
Fatwa
Mike
Aug 8, 2016, 12:27:45 PM8/8/16
to Genome Engineering using CRISPR/Cas Systems
Hi Fatwa,
very interesting to hear your perspective. I've never seen a low R2 from TIDE before, but looking at the TIDE publication there is a section at the end that says
"Decomposition results with a low R2 must be interpreted with caution. As a rule of thumb, we recommend to aim for a background signal of aberrant sequences before the break site <10% (both control and test sample), and R2 > 0.9 for the decomposition result."
I'm not totally sure what would give a poor R2 value, do you see high levels of aberrant sequences in your wild type samples in these cases?
Cheers,
Mike
very interesting to hear your perspective. I've never seen a low R2 from TIDE before, but looking at the TIDE publication there is a section at the end that says
"Decomposition results with a low R2 must be interpreted with caution. As a rule of thumb, we recommend to aim for a background signal of aberrant sequences before the break site <10% (both control and test sample), and R2 > 0.9 for the decomposition result."
I'm not totally sure what would give a poor R2 value, do you see high levels of aberrant sequences in your wild type samples in these cases?
Cheers,
Mike
Mike
Aug 9, 2016, 10:15:22 AM8/9/16
to Genome Engineering using CRISPR/Cas Systems
Hi,
I also saw this section on the webtool troubleshooting about low R2:
---
Sometimes the quality of the peaks in chromatogram looks fine, but the file has some wrongly unannotated or wrongly additional annotated nucleotides. These will interfere with the indel spectra (see figure wrongly unannotated nucleotide). TIDE gives a warning when the spacing between the nucleotides in the chromatogram of the sequence trace are not consistent, which is often an indication for wrongly unannotated or wrongly additional annotated nucleotides. Then the sequence file cannot be used for a reliable TIDE analysis. Try if possible to set the right boundary of the decomposition window lower. In case this warning stays, carefully investigate your chromatogram.
Poor sequence quality
Poor sequence quality can be observed in chromatogram (see figure Poor sequence quality). There is more noise present in the data with results in a lower R2.
----
So in general I think your idea that "TIDE-calculated efficiency/R-squared x 100%" is probably not the case and you need to adjust the settings and/or possibly obtain new sequencing trace files if R2 is low.
If you would like, send me an email directly with your sequence traces that give low R2 (and the guide sequence too) and I'll see if I can troubleshoot?
Cheers,
Mike
I also saw this section on the webtool troubleshooting about low R2:
---
Low R2 value
A low R2 can be caused when the settings are not optimal or when the sequence quality is not good.
Settings
By default, the decomposition window is set to its maximum size and the
Indel size range is set to 10. The settings can be adjusted in advanced
settings.
- Large indels are present in the sample. By default the decomposition is calculated with a maximal indel size of 10. When larger indels are present, they will cannot be modeled, which will result in a low R2. Try to increase the Indel size range to test if this improves the fit (see figure Indel size range)
- Poor local quality of the sequence trace. Often the end of the sequence is of low quality. This can be observed in the quality plot that shows a high aberrant sequence signal at the end of the sequence trace (see figure Poor quality sequence end). This compromises the decomposition of the sequence trace. Adjust the boundaries of the decomposition window in such a way that it will not overlap with the region that is of low quality.
- Repetitive regions in the sequence trace. These regions can be observed in the quality plot as a sudden stretch without aberrant nucleotides (see figure Repetitive region). This region might interfere with the decomposition of the sequence trace. Adjust the boundaries of the decomposition window in such a way that it will not overlap with the repetitive sequence part.
Sometimes the quality of the peaks in chromatogram looks fine, but the file has some wrongly unannotated or wrongly additional annotated nucleotides. These will interfere with the indel spectra (see figure wrongly unannotated nucleotide). TIDE gives a warning when the spacing between the nucleotides in the chromatogram of the sequence trace are not consistent, which is often an indication for wrongly unannotated or wrongly additional annotated nucleotides. Then the sequence file cannot be used for a reliable TIDE analysis. Try if possible to set the right boundary of the decomposition window lower. In case this warning stays, carefully investigate your chromatogram.
Poor sequence quality
Poor sequence quality can be observed in chromatogram (see figure Poor sequence quality). There is more noise present in the data with results in a lower R2.
----
So in general I think your idea that "TIDE-calculated efficiency/R-squared x 100%" is probably not the case and you need to adjust the settings and/or possibly obtain new sequencing trace files if R2 is low.
If you would like, send me an email directly with your sequence traces that give low R2 (and the guide sequence too) and I'll see if I can troubleshoot?
Cheers,
Mike
Fatwa Adikusuma
Aug 10, 2016, 12:04:33 AM8/10/16
to Genome Engineering using CRISPR/Cas Systems
Hi Mike,
Thank you for your response. Indeed it depends on the sequencing quality. One day I got low R2 values, another day I got a good one. I have played around with the advance setting to get the best R2 value. However, even with a good R2 value (0.9), TIDE would not tell you that your actual efficiency is close to 100% (based on more thorough analysis such as deep sequencing). And getting a perfect R2 probably requires tremendous effort.
Cheers,
Fatwa
jean-pierre de villartay
Aug 10, 2016, 3:20:05 AM8/10/16
to Genome Engineering using CRISPR/Cas Systems
Hi Fatwa (and the others)
as discussed by Bas van Steensel in his original paper the mutagenesis score calculated by TIDE cannot exceed 75% since in 25% of cases the same base as the original will be incorporated no matter what. If you think about it (please correct me if I am wrong) the real mutagenesis score is 4/3 that of calculated by TIDE. This means that the score is somehow under-evaluated, especially for high scores. Indeed 60% score is really 80% mutagenesis. The bad news is that this calculation will not turn a bad score into a good one: 10% score correspond to only 13% mutagenesis.
Hope this help.
jp
as discussed by Bas van Steensel in his original paper the mutagenesis score calculated by TIDE cannot exceed 75% since in 25% of cases the same base as the original will be incorporated no matter what. If you think about it (please correct me if I am wrong) the real mutagenesis score is 4/3 that of calculated by TIDE. This means that the score is somehow under-evaluated, especially for high scores. Indeed 60% score is really 80% mutagenesis. The bad news is that this calculation will not turn a bad score into a good one: 10% score correspond to only 13% mutagenesis.
Hope this help.
jp
Mike
Aug 10, 2016, 9:12:40 AM8/10/16
to Genome Engineering using CRISPR/Cas Systems
Hi JP,
I have written to Bas van Steensel to ask for his input here to clarify for certain, but I think you are incorrect about multiplying your score by 4/3. For starters, I've had many % total eff. scores over 90%, so you cannot multiply this by 4/3 or it would result in >100%. Also, I think this part of the paper refers to the the aberrant sequence graph, not the final % total efficiency score. It is explaining that if you had a 100% targeted sequence, the AVERAGE of all the light green bars would = 75%, not 100%. This is for the reason that you described, 25% chance of the changed nucleotide being the same as wildtype, but of course you will have a distribution of bar heights, from 100% to 0% in 25% steps.
In any case, this is all accounted for by the software when calculating the final % total eff. score, this score does not need to be scaled, you should take it as given. The only thing I find is that it probably underestimates the efficiency for very highly targeted samples (eg: monoclonal lines, polyclonal lines targeted by constitutive lentivirus etc.), I have never seen higher than about 96%, even for samples that are completely targeted if I look by western blot (no signal even after very long exposure where WT band is extremely overexposed, if 4% protein remained I would have seen it). My guess is that the inherent noise in chromatogram signal (which is what leads to a low level of aberrant sequence in the WT sample) means you can never really approach 100%. So I would guess any % total efficiency score should be interpreted at +/- 5%? But either way it gives a pretty accurate picture, probably at least as accurate as other techniques, and more accurate than most (except NGS or WB).
Hope this helps, and hope one of the original authors can reply here.
Mike
I have written to Bas van Steensel to ask for his input here to clarify for certain, but I think you are incorrect about multiplying your score by 4/3. For starters, I've had many % total eff. scores over 90%, so you cannot multiply this by 4/3 or it would result in >100%. Also, I think this part of the paper refers to the the aberrant sequence graph, not the final % total efficiency score. It is explaining that if you had a 100% targeted sequence, the AVERAGE of all the light green bars would = 75%, not 100%. This is for the reason that you described, 25% chance of the changed nucleotide being the same as wildtype, but of course you will have a distribution of bar heights, from 100% to 0% in 25% steps.
In any case, this is all accounted for by the software when calculating the final % total eff. score, this score does not need to be scaled, you should take it as given. The only thing I find is that it probably underestimates the efficiency for very highly targeted samples (eg: monoclonal lines, polyclonal lines targeted by constitutive lentivirus etc.), I have never seen higher than about 96%, even for samples that are completely targeted if I look by western blot (no signal even after very long exposure where WT band is extremely overexposed, if 4% protein remained I would have seen it). My guess is that the inherent noise in chromatogram signal (which is what leads to a low level of aberrant sequence in the WT sample) means you can never really approach 100%. So I would guess any % total efficiency score should be interpreted at +/- 5%? But either way it gives a pretty accurate picture, probably at least as accurate as other techniques, and more accurate than most (except NGS or WB).
Hope this helps, and hope one of the original authors can reply here.
Mike
jean-pierre de villartay
Aug 10, 2016, 9:29:46 AM8/10/16
to Genome Engineering using CRISPR/Cas Systems
Thanks Mike for this update. I had not realized that TIDE already make some sort of corrections.
Then I found troubling that in my case, the complete disappearance of a SacI site on gel only translates into 70% mutagenesis.
best
jp
Then I found troubling that in my case, the complete disappearance of a SacI site on gel only translates into 70% mutagenesis.
best
jp
Message has been deleted
Eduarde Rohner
May 16, 2017, 8:07:37 AM5/16/17
to Genome Engineering using CRISPR/Cas Systems
Hello,
Hopefully someone is still around to answer =P
I was wondering how to interepret difference is efficency (2-15% in my case) between FWD and REV sequences of the same sample. The overall spectra of mutations is identical but the efficencies can differ (I usually get higher values with the REV sequencing).
The main difference between both is the decomposition sequence (that was shortend in the REV sequence (still 250bp long, instead of >400bp for the FWD).
The overall quality is very high and I have R^2 values >0.96 in all samples (FWD and REV). And also to mention my gRNA sequence is on the REV strand in all cases.
Best
Edu
Hopefully someone is still around to answer =P
I was wondering how to interepret difference is efficency (2-15% in my case) between FWD and REV sequences of the same sample. The overall spectra of mutations is identical but the efficencies can differ (I usually get higher values with the REV sequencing).
The main difference between both is the decomposition sequence (that was shortend in the REV sequence (still 250bp long, instead of >400bp for the FWD).
The overall quality is very high and I have R^2 values >0.96 in all samples (FWD and REV). And also to mention my gRNA sequence is on the REV strand in all cases.
Best
Edu
Reply all
Reply to author
Forward
0 new messages