[an error occurred while processing this directive]
On this page:
 

TiP

This page describes the proprietary FTDNA TMRCA calculator, known as TiP. The acronym stands for "Time Predictor".

A TMRCA (time-to-most-recent-common-ancestor) calculator compares a pair of Y-DNA STR haplotypes and estimates the cumulative probabilities, by generation, for the time when a possible common ancestor lived. We note that some are skeptical of all TMRCA calculators; we, however, believe they are necessary when used properly.

How TiP Differs

A variety of TMRCA calculators exists. Most rely on an overall average of STR marker mutation frequencies. For example, the McDonald and Genebase calculators have the user input the following:

We might regard the output as a "first-order approximation" of TMRCA. The limited input  and calculation algorithms fall far short of reflecting what happens genetically. Such rudimentary calculators have at least two problems:

  1. They inadequately model marker mutation behavior. They assume an "infinite alleles" model for mutations, in which a marker is free to change by any amount in a single mutation.  More recent research shows that only some markers follow the infinite alleles model. Most markers follow a "stepwise" model, in which each mutation produces a single step (up or down) of genetic distance.
  2. They do not account for individual differences in marker mutation frequencies. In fact, the range of frequencies is quite wide -- from one per 28 for CDY to one per 11,111 for DYS426.

TiP, on the other hand, overcomes these problems:

  1. TiP uses a stepwise model for most markers and an infinite alleles model for those markers known to behave in that fashion.
  2. TiP uses individual marker mutation rates.
Some have observed that other events (e.g., RecLOH) which can affect many markers are not well-handled by TiP.

These features make TiP, in the author's opinion, the most accurate TMRCA calculator currently available. We regard it as the best present measure of the similarity (or dissimilarity) of a pair of haplotypes.

How TiP Works

TiP is available on the the FTDNA website (www.familytreedna.com) for FTDNA clients with Y-STR results. It compares two sets of results, as stored in the FTDNA database, and displays calculated cumulative probabilities by number of generations.

Use of TiP Scores

We refer to the 37-marker, 24-generation, no paper-trail-adjusted TiP-derived cumulative probability simply as a "TiP score" and typically use it for grouping purposes. Two or more members with such scores >= 80% are assigned to a Taylor genetic family.

When 67- or 111-marker TiP scores are possible, they may over-ride a negative grouping determination at 37 markers.

Paper Trail Adjustment

Increasing this number from the default value of 1 shifts the probability curve to the right by that number of generations. Because no more than 24 generations will be displayed, we have found it more useful to make the adjustment (if desired) after the report is displayed, rather than supply it as input to the algorithm.

TiP versions

There have been at least three versions of TiP, perhaps more. One version (TiP v1) was in use until 12 Dec 2012, at which point another version (TiP v2) was implemented. Observed problems led to another version (TiP v3) beginning about February 2013.

TiP Performance

We have found TiP to discriminate well between "good matches" and "bad". TiP scores tend to cluster near the high and low ends of a probability scale.

In August & September 2011, Taylor Family Genes conducted a review of TiP, calculating TiP scores for all 240 members who had at that time tested 37 markers. Each member's TiP scores against all other members were calculated. This resulted in a matrix of 57,120 TiP scores, ranging from 0% to 100%.

The chart on the left shows the overall distribution of TiP scores in the Taylor project. Note that 80% of all scores are less than 10% cumulative probability and 90% less than 20% p. Less than 3% of scores are in the "marginal" categories of 40% to 79% p.

Fewer than 1% of all TiP scores reach the 80% at 24 generations threshold for inclusion in a genetic family.

The chart to the right shows the distribution of the same TiP scores by haplogroup.

An anomaly is the J2 peak at 30-39%. It may be a reflection of few J2 project members. Only four have tested 37 markers.

Methodological Notes: The data is solely intra-haplogroup; cross-haplogroup calculations were not included. (See Cross-haplogroup.) Within haplogroups, the analysis counted every pair of members twice, once for each. This produces twice the number of scores but is not a problem when using percentages. The score for member A vs. B will be the same as that for B vs. A.

Cross-haplogroup scores would have been almost invariably zero.

R1b Performance

The largest haplogroup in the project (70% of members) is R1b1a2 (R-M269). This haplogroup includes two 12-marker haplotypes (WAMH & Niall) whose bearers often have hundreds of reported "close matches" at 37 markers. What is TiP performance for this group?

The graph on the left shows the complete distribution of 28,559 TiP scores for 169x169 R1b members, of whom 98% are R1b1a2. The graph on the right shows that part of the distribution for p>30% so that the scale may be expanded.

Cross-Haplogroup Performance

No systematic evaluation of TiP's performance when haplogroups differed was performed. However, a sample of 20 cross-haplogroup calculations were performed. All yielded TiP 37/24 scores of 0%, with mismatches ranging from 17 to 40.