SliderII: High Quality SNP Calling Using Illumina Data at Shallow Coverage

[pengchy]

Is SliderII's paper published?
And the picture in this link can not be displayed

http://www.bcgsc.ca/platform/bioinfo/software/SliderII

[nmalhis]

Hi Nix,

I used the Ensembl Variation database (version 50) SNPs.
You need to adjust the format.

Nawar

[Nix]

What coordinate system are you using for generating a table of known snps to feed into SliderII. Is this 1 based? or 0 based? Does anyone have a table for mouse 2007, mm9?

[nmalhis]

> Hi,
>
> I am very interested in your SNP Caller SilderII. I am trying to use it. I have one question for you. What's the meaning of SNP_in in the config file? I didn't find the explanation for the item from sliderII website.
>
> Thank you very much.
>
> Rebecca

SNP_in is the expected number of bases in the reference genome for each one SNP, for the human genome, this number should be 1000.

Nawar

[lh3]

2. Do you mean you exclude SNPs towards the ends of a read? These are the false SNPs caused by indels. A better strategy would be to filter out SNPs close to predicted indels.

3. Sorry that I did not read through the whole page. I now realize that this is a seeding-extension algorithm. You allow maximum one mutation in the seed but may extend the seed to allow more. By the way, the page said "the smaller the seed size is, the faster the alignment will be". Is this a typo?

[nmalhis]

- Regarding paralogous, Slider identify paralogous SNPs (and contig edge SNPs) as they are likely to be at the edges of the reads.
- Yes, Slider (and SliderII) allows up to one mutation, plus, it consider all possible bases in prb data, and when using PET reads, SliderII force align reads if other side is aligned.

Nawar

[nmalhis]

Yes, just check the link:

http://www.bcgsc.ca/platform/bioinfo/software/SliderII

N.

[bioinfosm]

Any insight on how slider results compare to MAQ SNP calling on single/paired data?

Thanks for this. I am always fancinated by slider. I guess this is the first SNP caller that explicitly use four quality values. James Bonfield and Mark Daly both believe and show some preliminary result that using four values leads to better SNP calls. Some comments on the figures at your website:

1. It is interesting to see you also come to the point of using known allele frequency as a prior, the same as BGI's SNP caller. When I did SNP calling for that NA18507, I also suggested this, but all the rest of people said it is cheating somehow and rejected my suggestion. They more like to think there are two problems: SNP discovery and genotyping. For SNP discovery, we only use a flat prior and for genotyping, we use the allele frequency.

2. How Slider detect paralogous regions? To detect CNV first and then filter out the SNPs in CNVs? I agree that setting maximum depth as is used by maq is not a good way.

3. I am not sure if I read your paper properly. As I understand, only one mutation (not sequencing errors) is allowed on one read. Is that right?

[bosTau2]

step by step

I checked http://www.bcgsc.ca/platform/bioinfo/software/SliderII
and think it does alignment by steps.

# Alignment.Java: Find read locations on the reference sequence with an exact match and one-off match (one base mismatch) to prb derived sequences.
# Extend.java: Expand reads to include up to 3 mismatches

[lh3]

Thanks for this. I am always fancinated by slider. I guess this is the first SNP caller that explicitly use four quality values. James Bonfield and Mark Daly both believe and show some preliminary result that using four values leads to better SNP calls. Some comments on the figures at your website:

1. It is interesting to see you also come to the point of using known allele frequency as a prior, the same as BGI's SNP caller. When I did SNP calling for that NA18507, I also suggested this, but all the rest of people said it is cheating somehow and rejected my suggestion. They more like to think there are two problems: SNP discovery and genotyping. For SNP discovery, we only use a flat prior and for genotyping, we use the allele frequency.

2. How Slider detect paralogous regions? To detect CNV first and then filter out the SNPs in CNVs? I agree that setting maximum depth as is used by maq is not a good way.

3. I am not sure if I read your paper properly. As I understand, only one mutation (not sequencing errors) is allowed on one read. Is that right?

[nmalhis]

SliderII is now available from:

http://www.bcgsc.ca/platform/bioinfo/software/SliderII

Sorry for the delay,

Nawar