Minor Allele Frequency Cutoff/Threshold

BertieWooster · 07-28-2009, 03:21 PM

G'day,

I am writing to enquire what maf cutoff/threshold people use when getting snps using the maq snp caller or any other means.

A lot of researchers seem to use 0.05 as applied for HapMap, but this is applicable for population data. Is there a "statistical" way to estimate this from the data or should one just do it arbitrarily?

Cheers!
BertieWooster

jaspersaris · 09-28-2009, 02:50 PM

Hai Bertie,
I am not sure what you mean. In a normal, autosomal, say recessive disease like situation I place my cutoff at 20%. That's based on Sangerseq based views and clean sequences in Seqscape. In situations were we aware of possible mosaic situations we first improve our sequences and then can pickup at 5-10%. So 0.05 in a none optimised sequenceproject is probable generating false-positives. I haven't explored/realised it yet, but yes, by counting each variant per base and having enough seq-depth would yield a cluster and borders there-off to estimate a general or floating statistics based cutoff.
Regards, Jasper

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Allele frequency calculation in SNP calling	Rachelly	Bioinformatics	5	05-14-2012 12:07 PM
Question on Minor allele Frequency in dbSNP135	ngarg	General	0	02-06-2012 10:55 AM
allele frequency threshold GATK	m_elena_bioinfo	Introductions	3	12-12-2011 12:29 PM
how to fetch the snp allele frequency?	dzmtnvmt	Bioinformatics	3	06-21-2011 04:44 AM
Connection between sequencing error and observed allele frequency	tcag	Introductions	0	03-21-2011 11:58 AM

07-28-2009, 03:21 PM	#1
BertieWooster Member Location: Sydney, Australia Join Date: Jul 2009 Posts: 14	Minor Allele Frequency Cutoff/Threshold G'day, I am writing to enquire what maf cutoff/threshold people use when getting snps using the maq snp caller or any other means. A lot of researchers seem to use 0.05 as applied for HapMap, but this is applicable for population data. Is there a "statistical" way to estimate this from the data or should one just do it arbitrarily? Cheers! BertieWooster

09-28-2009, 02:50 PM	#2
jaspersaris Member Location: NL Join Date: Nov 2008 Posts: 6	Hai Bertie, I am not sure what you mean. In a normal, autosomal, say recessive disease like situation I place my cutoff at 20%. That's based on Sangerseq based views and clean sequences in Seqscape. In situations were we aware of possible mosaic situations we first improve our sequences and then can pickup at 5-10%. So 0.05 in a none optimised sequenceproject is probable generating false-positives. I haven't explored/realised it yet, but yes, by counting each variant per base and having enough seq-depth would yield a cluster and borders there-off to estimate a general or floating statistics based cutoff. Regards, Jasper