Archiver > GENEALOGY-DNA > 2003-11 > 1068496420

Subject: Re: [DNA] Measurement of fragment lengths
Date: Mon, 10 Nov 2003 15:33:47 -0500 (EST)
References: <001101c3a74c$df744d00$3ca3fea9@NancySonyIV>
In-Reply-To: <001101c3a74c$df744d00$3ca3fea9@NancySonyIV> (

Nancy wrote:
> It could be the result of a variant allele. A variant allele is an STR
> allele with an incomplete repeat. For example a 9.3 would mean 9
> repeats of a four nucleotide "motif" plus 3 nucleotides of a tenth
> repeat where one nucleotide has been lost by deletion (a kind of
> mutation.) I don't know what an 11.46 or 11.44 would represent.

I think the point that Malcolm was making includes the notion of
measurement error. There is, in fact, a trade-off between speed
and accuracy in the measurement process. We would expect the labs
to do whatever is necessary to keep the error level down to +/-
0.1 or so (standard deviation), but it would be crazy to work hard
getting the accuracy any tighter than that. Consequently, there
would occasionally be "outlier" measurements off by more than one
or two standard deviations. Being off by 0.46, however, is too
much to allow. In any case, the automated allele calling should
flag these extreme cases as errors, requiring that the measurement
be done again. Note that the random component of measurement error
tends to average out with repetition.

John Chandler

This thread: