9919andMultiRecLOH

  • 214 members

About us

Welcome to the 9919 and RecLOH Project

What is a recLOH?

A recLOH, or "Recombinational Loss of Heterozygosity", in plain English, is when one piece of DNA overwrites another piece.

Both copies become identical and lose their heterozygosity - that is, they lose their differences from each other - and the result is a doubled marker value.

See here for much more detailed, scientific descriptions:

http://en.wikipedia.org/wiki/RecLOH
http://www.dna-fingerprint.com/modules.php?op=modload&name=Sections&file=index&req=viewarticle&artid=9
http://www.dna-fingerprint.com/static/MakingSenseOutOfYPolymorphismsMPI2011.pdf                         <----- This page takes a long time to load.

Here is an excellent diagram of the palindromic multicopy markers, copyright 2006 - Thomas Krahn:
http://www.dna-fingerprint.com/static/PalindromicRegion-V3.pdf

See http://dna-project.clan-donald-usa.org/DNAdna3.htm, explaining how 464=13-14-16-18 "could become 16-16-18-18 in one event, not seven".



You can see evidence of a recLOH event in the six multicopy palindromic markers 385, 464, YCAII, CDY, 395s1, and 413 from FTDNA's standard 67-marker test, and in several other markers that you can only see results for in the special-order-only "Panel 5 Palindromic Pack" (see the end of this page for more details on P5PP).






What are "multicopy" markers?

Multicopy markers are genetic markers that occur more than once, and have a dash separating two or more values, such as 385=11-13 or 464=15-16-16-17.  There are no multicopy markers in the 68 to 111 range.  The multicopy markers all occur in the marker # 1 to # 67 range.  The multicopy markers in your standard 67-marker STR set are these eight markers:

     385
          389
     459
     464
     YCAII
     CDY
     395s1
     413

Seven of these - all except marker 389 - occur in the "palindromic" region of the Y-chromosome, a region where the DNA loops back on itself several times.  RecLOHs can involve these 7 palindromic markers, whereas 389 is not involved in recLOH events.

Typically, the multicopy markers have two separate readings, one for each of two alleles.  The exceptions are:
- Marker 464, which typically has 4 alleles, but can have 5, 6, or even more (up to 12, according to FTDNA).
- Marker 389, which has only one amplified location on the Y chromosome, with one reading representing one allele when read in the forward direction, and another reading representing BOTH alleles when read in the reverse direction.  For more details, see "What is special about marker 389" at the end of this page.






What do recLOHs look like?

An example of a recLOH event could be this combination consisting of three likely recLOHs:

459=9-9
464=16-16-17-17
CDY=36-36

In this example, assuming a recLOH event did indeed occur in one of the person's ancestors (or perhaps even in the person himself):
- The 9-repeat section of the 459 marker overwrote the other half.  The pre-recLOH value may have been 459=9-10.
- The 36-repeat half of the CDY marker overwrote the other half.  The pre-recLOH value may have been CDY=36-38.
- The "16-17" half of the four-part 464 marker overwrote the other half, resulting in "16-16-17-17" (since 464 is always stated from lowest values to highest values).
The pre-recLOH ancestor may have had 15-15-16-17, or 15-16-16-17, or 15-16-17-18, or something entirely different.  Unless the person has close matches who do not have the recLOH at the 464 marker, we really have no way of knowing what the pre-recLOH 464 values were.  All we know is that a chunk with "16-17" overwrote the other half.

If a person has close matches who do not have the recLOHs, then we can reasonably predict what the pre-recLOH values were.  For example, if a person's close matches have 464=15-15-16-17 and the person has 464=16-16-17-17, then we can assume that sometime in the past few generations in this man's lineage, there was a recLOH event changing the 464 values from 15-15-16-17 to 16-16-17-17.

There are an infinite number of possibilities of what recLOHs could look like.  The multicopy markers have different mutation rates and different positions on the Y chromosome's palindromic region, so some of them are more susceptible to recLOH events than others.  Some particular recLOH combinations, such as the combination of 459=9-9 and YCAII=19-19, are apparently much more common than other combinations.










If a kit only has one doubled multicopy marker, is it a recLOH or is it a normal "step" mutation?

If a person only has one doubled multicopy marker, such as 459=9-9, then we have no way of knowing (without additional tests) whether this is a normal one-step mutation from 459b=10 to 459b=9, or instead is the result of a recLOH event in which 459a=9 overwrote 459b=10.  When a person has two or more doubled multicopy markers, especially if the doubled values are relatively unusual, then the chances that there was a recLOH event greatly increase.  We have no way of knowing if a major jump from, say, 385=10-15 to 385=15-15 was the result of one recLOH event, or the result of a massive five-step mutation, but a recLOH would be the more logical explanation for a change of this magnitude.

From outward appearances, a recLOH in very slow mutating 395s1 alone would look indistinguishable from a normal one-step or two-step mutation.  We could only deduce that it was a recLOH if it existed in concert with other doubled palindromic markers.










If a kit has two or more doubled multicopy markers, were they the result of one recLOH event, or multiple recLOH events?

We also have no way of knowing if the two or more doubled multicopy markers were the result of one recLOH event, or the result of multiple events.  Here we should trust Occam's Razor:  The simplest explanation is usually the correct one.  Chances are, one recLOH event affected multiple markers in one fell swoop.










How stable are recLOHs compared to individual STRs, and compared to SNPs?

The recLOH's stability probably depends on at least two factors:
(1) The mutation rate of the marker in the recLOH configuration,
(2) The palindromic position of the marker on the Y chromosome.

If there were two or more recLOHs resulting from a recLOH event, the recLOHs are unlikely to remain intact as a coherent unit.  After a recLOH event, the individual recLOHs can mutate, even mutating back to the original condition.  For example, take a look at the STR set for N64980 (Chandler).  You can see his STRs here:
http://www.familytreedna.com/public/1610johnchandlerdna/default.aspx?section=yresults

He apparently had a back mutation from 459=9-9 to 9-10 in his lineage.  The first mutation was likely a recLOH that happened hundreds or even thousands of years ago in the common ancestor of all of the z9919 men, from 459=9-10, YCAII=19-23 to 459=9-9, YCAII=19-19.  The second mutation in N64980's lineage was likely a normal one-step upward mutation from 459b=9 to 459b=10.  His YCAII stayed at 19-19, and his 640 still has the unusual value of 12, confirming he belongs in z9919-A.










How common are recLOHs?

RecLOHs are unusual but not rare.  They were once regarded as rare, and as such, the L21+ 9919 men were grouped together.  There were some who questioned whether they should be grouped together at all, as they might just be random haplogroups who happened to have the same recLOH.  Later we discovered that almost all of them do have other persistent off-modals, most prominent of which is 640=12 in the "A" groups, and the 9919 haplotypes now cluster in several known signatures.  Besides seeing these signatures in the R1b-L21_Haplotypes spreadsheet on L21 Yahoo Group, you can also see a summary on the RESULTS tab of this project.

You can find hundreds of examples of kits with likely recLOHs on the website called Semargl.  Some combinations of likely recLOHs are quite common, whereas other combinations are exceptionally rare.  To demonstrate for haplogroup R1b, go to Semargl's "query to the website database", http://www.semargl.me/en/dna/ydna/query, select Haplogroup R1b, and type in two unusual doubled or quadrupled values such as:


385=11-11, 459=10-10, and 395s1=16-16

or

459=9-9 and YCA=21-21

or

459=10-10 and YCA=20-20

or

YCA=19-19 and 395s1=15-15

or

385=13-13 and 459=9-9

or

459=9-9 and 464=15-15-16-16-17-17

or

456=15-15-18-18 and YCA=23-23

or

464=15-15-16-16 and 413=22-22
  
or

459=9-9 and 464=16-16-16-16

or

459=10-10 and YCA=19-19

The last example results in a kit with several doubled markers including CDY 39-39, 395s1=16-16, and six 464 values, 17-17-17-17-18-18.

Some combinations on Semargl produce few or no results.  For example, {385=14-14, YCAII=23-23} produces no results.

The combinations 385=11-11 and 385=14-14 each tend to occur alone; that is, without any other doubled multicopy markers.










What is an example of the most mega-hyper-fantastic recLOH event that could possibly occur?!

Here is an example of a fictitious kit that has seven recLOHs, including a recLOH in a rare sextuple-copy 464 marker:

13 24 14 11 14-14 12 12 12 13 13 29 17 9-9 11 11 25 15 19 30 15-15-17-17-18-18 11 11 19-19 16 15 18 17 41-41 12 12 11 9 16-16 9 10 10 8 10 10 12 25-25

These recLOH values are all unusual, but the rest of the markers are perfectly normal for an R1b-P312 person.  In this extreme case, the person would have few or no matches.  If he only had two or three recLOHs, then he'd probably still have dozens of matches, but because of the recLOHs, his matches would seem genetically more distant than they really are.

For comparison, here is L21 modal, up to the last multicopy marker, 413:

13 24 14 11 11-14 12 12 12 13 13 29 17 9-10 11 11 25 15 19 30 15-15-17-17 11 11 19-23 16 15 18 17 37-38 12 12 11 9 15-16 9 10 10 8 10 10 12 23-23










If a person has doubled multicopy markers and few or no matches at 12, 25, 37, 67, or 111 markers, should I suspect that these doubled markers are recLOHs?

No.  You need to look at the doubled markers and determine how likely they are to be recLOHs.  Are they normal doubled values, such as 413=23-23, or are they unusual doubled values?

Whether doubled markers are recLOHs and whether a person has few or no matches don't have much to do with each other.  RecLOHs cause a person to have *fewer* matches, as recLOHs are typically unusual values, but recLOHs do not cause a person to necessarily have *few* matches.

On one extreme, a person could match L21 modal except for a few recLOHs, such as 385=11-11 and YCAII=23-23.  Even with these recLOHs, he would still probably have hundreds of matches.  On the other extreme, a person could have a dozen other off-modals, besides the recLOHs.  He would probably have few or no matches, but his recLOHs would be only one factor of many in why he had so few matches.










How do you measure GD when there's a recLOH?

Measuring GDs for recLOHs is tricky.  If a person has a recLOH of 385=11-11, and his closest matches all have 385=11-14, you treat the change from 14 to 11 as GD of 1, rather than 3.










How do recLOHs affect GD?

RecLOHs can make a person seem further away from his matches than he really is.  If a person has two recLOHs that caused a 3-step change on one marker and a 2-step change on another marker, such as 385=14-14 and CDY=34-34, where his closest matches are 384=11-14, CDY=34-36, then his GD will seem to be 5 but it is really 2.










What kind of 464 values would suggest a recLOH event has occurred?

The 464 marker, a multicopy marker that normally has 4 values, on rare occasions can have 5, 6, or even more values.

According to Butler (2003, NIST paper), "Analysis of the directionality of DYS464 sequences along the Y-chromosome indicates that it is really a duplicated duplicate locus rather than an independently quadruplicated one."

Thus, if you see a 464 value that looks like one half of the 464 quad overwrote the other half, producing a double-double (such as 16-16-17-17) or a single-value quadruple (such as 15-15-15-15), this is likely a recLOH.

Before:

15-15-16-17

After:

15-15-15-15 (the chunk "15-15" overwrote the chunk "16-17")
OR
15-15-16-16 (the chunk "15-16" overwrote the chunk "15-17") - since 464 values are always stated from lowest to highest, you report the two 15s first and two 16s last.
OR
16-16-17-17 (the chunk "16-17" overwrote the chunk "15-15") - since 464 values are always stated from lowest to highest, you report the two 16s first and two 17s last.

The value "15-15-17-17" *could* be a recLOH, but it is also the most common R1b P312 and L21 value, so it is highly unlikely to be a recLOH, at least in an R1b person.

On the other hand, 464 triple values could also represent one or more recLOH events.  The process by which R1b took on the cccg configuration of 464, while most other haplogroups have gggg, probably involved an initial mutation and two successive recLOH events (see John McEwan's paper, June 2006).  Rozhanskii and Klyosov (2009) describe a change from 15-15-17-17 to 15-17-17-17 as a recLOH event, and count it as one mutation.

If you see a triple value, such as 16-16-16-17, you need to know the base haplotype (the original haplotype before a mutation resulted in the new haplotype 16-16-16-17) to gauge the chances that a recLOH event occurred.  If the base haplotype is 15-16-16-17, there may have been a recLOH event, with one of the two loci with 16 overwriting the locus with 15, or there may have been a simple one-step mutation at the locus with 15 mutating to 16.  There's really no way to know unless the 464x configuration also changed.  If there is a two-count change, for example, from 14-16-16-17 to 16-16-16-17, or from 16-16-17-18 to 16-16-16-17, then the chances that there was a recLOH event are much greater.









Coud a 464 value such as 15-15-15-15-16-16 or 15-15-16-16-17-17 represent a recLOH?

Yes.  There are two ways to think about this - two ways of phrasing the same idea:
(1) Each copy has a complement on the other half of the palindrome.  In this case, you could say that each of 3 markers overwrote each of three other markers.
(2) One half of the 464 marker completely overwrote the other half.  In this case, the 3-copy chunk that did so would be 15-15-16 or 15-16-17, respectively.

For example, let's assume that the pre-recLOH value was 464=13-15-15-16-17-18.  On one strand of the DNA in the palindromic region of the Y chromosome, you might have 13 repeats, 17 repeats, and 18 repeats.  On the complementary strand, you might have 15 repeats, 16 repeats, and 15 repeats.  If there were a major deletion or DNA replication error in the first strand, the DNA would attempt to repair itself by copying the second strand onto the first strand.  This would result in 15, 16, and 15 on one strand and also 15, 16, and 15 on the complementary strand.  When a lab measured the values, the lab would report them from lowest to highest, and would therefore report 15-15-15-15-16-16.

Similarly, if the second strand had 16 repeats, 15 repeats, and 17 repeats, and overwrote the complementary strand, this would result in 16, 15, 17 on one strand and also 16, 15, 17 on the complementary strand.  Stated lowest to highest, the lab would report this as 15-15-16-16-17-17. 










Thomas Krahn's diagram shows 464 occupying two palindrome arms, P1 and P2, with two copies on each arm.  Occasionally, the 464 marker has more than 4 copies.  Where would the fifth, sixth, or seventh copies exist?  Would they exist on P1, on P2, or somewhere else altogether?

I have no clue.     <----- If you know the answer to this, please send me an email.










Can the 464 marker be a recLOH if it has an odd number of copies, such as 5 or 7 values instead of the usual 4?

I don't know the answer.  I'm guessing no, since you couldn't explain the 5th or 7th value as being the product of a recLOH, with no complementary 6th or 8th value overwriting anything.  However, an odd-copy 464 could be a mutation from an earlier recLOH condition.  For example, 16-16-17-17 could mutate to 16-16-17-17-17.










How can I judge more accurately whether one or more recLOHs exist?

Judging recLOHs is an art as much as a science.  There are several factors to consider, including the mutation rates and the palindromic locations of the particular markers that look like recLOHs, the person's haplogroup (both major, such as R1b1a2, and smaller, such as a specific branch of DF13), what genetic signature the person fits or seems to fit (such as z9919), and the general STR patterns of the person's closest matches.  RecLOHs are easiest to gauge when the doubled marker values are unusual, and the marker's mutation rate determines how far the marker can stray downward or upward before your recLOH "ears" perk up.

CDY is one of the fastest mutating markers and you can expect frequent mutations into a double configuration, such as 36-36, 37-37, 38-38, or 39-39, without involving recLOH events.  Common doubled values (such as 37-37) would be unlikely to indicate a recLOH.  If you see an unusually low CDY=34-34 or an unusually high CDY=41-41, this would suggest a good possibility of a recLOH, especially if it is in combination with other unusual doubled markers and especially if the man's closest matches have much different CDY values (such as 34-37 or 38-41).  Then again, there's always the possibility of a normal 3-step mutation from 37 to 34, or from 38 to 41.

At the other extreme, 395s1 is very slow mutating, with a mutation rate less than 0.001, and you would expect 395s1 to have low susceptibility to recLOHs.  Interestingly, despite 395s1 having a much slower mutation rate than 413, both markers seem to be equally susceptible to recLOHs.  This is based on my unscientific random sampling of results when searching for doubled marker combinations on Semargl.  Maybe this has something to do with 413's position, all alone on palindromic arm P8, as opposed to 395s1's more crowded position on the P5 palindrome.

The larger the increase or decrease in value, the easier you can confidently predict that a recLOH event has occurred.  Marker YCAII is the "platinum" marker in the R1b1a2 recLOH kingdom; the easiest marker to predict as a recLOH for R1b1a2, since it has a four-step gap between its copies at the modal values of 19-23.  Any recLOH that occurs in YCAII is likely to shift the value of one copy by a whopping 4 steps.  Marker 385 is the "gold" marker in the recLOH kingdom, with a huge 3-step gap between its copies at the modal values of 11-14.  Marker 464 is easy to predict as a recLOH as long as it is not at its modal doubled value 15-15-17-17.  However, 464 is fast mutating so you can expect that some doubled or quadrupled 464 values are normal step mutations that merely look like recLOHs, and you really need to see other doubled markers to clinch the deal.  Markers 459, CDY, and 395s1 are more difficult to predict as recLOHs, with only a one-step gap between their copies at the modal values for each marker, and 413 is the most difficult of all, as the modal values are doubled and the marker is slow mutating to boot.  One copy of 413 could overwrite the other copy and we would never know anything had changed.  In the recLOH kingdom, 413 is fool's gold.

To determine whether or not a recLOH has likely occurred, you need to know what the normal or "modal" values are for your haplogroup, and whether those normal values include any doubled values. 

Certain double marker values are common in certain haplogroups and may be super-stable, resisting downward or upward movement, or quickly mutating back to the most stable configuration.  Alternatively, certain haplogroups may have had recLOH events early in their formation, and these ancient recLOHs may have persisted in nearly everyone in that haplogroup to the current day.  Whatever the case may be, there are many, relatively common doubled marker values in various haplogroups that are probably not indicative of a recLOH, or at least not indicative of a recent, non-ancient recLOH.  For example:

In haplogroup G (most common in the Caucasus region and in Sardinia) and haplogroup O3 (common in East Asia), the most common value for 459 is 9-9.

In G's subclade G1a1, the most common value of YCAII is 20-20.

Haplogroup D (common in Tibet and Japan) almost always has 459=9-9 and often has a 464 quad consisting of only one or two values, such as one of these combinations:
14-14-15-15
16-16-16-16
16-16-17-17
17-17-17-17
16-16-18-18

In haplogroup I1, the doubled value 395s1=15-15 is quite common.

In haplogroup R1a1a, the doubled value 395s1=17-17 is quite common.


In R1b P312+ men (including branches L21, U152, DF27, and DF19), markers 464 and 413 are naturally doubled:


     464=15-15-17-17

     413=23-23


As with doubled markers common in other haplogroups, marker 464's and 413's naturally doubled configuration in R1b1a2 could indicate a recLOH event early in R1b1a2's history that carried forward into nearly all descendants, or could simply indicate that this doubled condition is the most stable configuration for these two markers.  In order to say that 464 or 413 were likely recLOHs, we would need to see unusual doubled values, such as 464=16-16-18-18 and 413=24-24.  The more uncommon the doubled values, and the more of them there are, the more likely that they are recLOHs.  Anytime you see YCAII=19-19 or YCAII=23-23 in a P312+ person whose closest matches have YCAII=19-23, the person's unusual 19-19 or 23-23 is almost certainly a recLOH, as the chances are slim that one copy mutated one step four times, or mutated four steps one time.  More likely is that a recLOH event occurred, overwriting one copy with the other.  The same goes for 385=11-11 or 385=14-14.  Then again, if a person's closest matches all have 385=11-12, then a change to 11-11 would more likely be a one-step mutation than a recLOH.

Case 1:
A person has 385=11-11.  He has several close matches who all have 385=11-12.
Conclusion 1:  There was a normal downward mutation from 385b=12 to 385b=11.  There was no prior recLOH event in 385.  There may have been normal step mutations over the centuries, from 385b=14 to 13, and then from 13 to 12.
Conclusion 2:  There was a recLOH event in this person's lineage, in which 385a=11 overwrote 385b=12.  From outward appearances, this looks just like a normal step mutation unless there are other unusual doubled markers.
Conclusion 3:  A recLOH occurred a long time ago, from, say, 11-14 to 11-11.  Then, a normal mutation occurred from 11-11 to 11-12.  The men with this mutation were more successful for whatever reason, and had many more children.  Then, the person in question had a back mutation in himself or in his lineage from 11-12 to 11-11.

Case 2:
A person has 385=11-11.  Most of his close matches also have 385=11-11, but a few have 385=11-12.  Conclusion:  A recLOH occurred long ago from unknown pre-recLOH values, probably 11-14, to the newer values of 11-11.  Since then, some of the men had normal upward mutations from 385b=11 to 385b=12.










What are the criteria for a likely recLOH, reduced down to the bare naked essence?

(1) More than one doubled multicopy marker (385, 464, YCA, CDY, 395s1, 413); that is, two, three, or more doubles.  The maximum number in a 67-marker kit would be 7.
(2) Uncommon values for the doubles.










Can you show me the general though process of looking with these two basic criteria in mind?

Sure, follow along.  These are all hypothetical examples.

This combination in a P312+ person would almost certainly consist of multiple recLOHs, as the values are doubled AND are relatively unusual:
385=14-14, 464=15-15-18-18, 413b=25-25.

This combination could consist of recLOHs in a P312+ person, but these values are more common, so the chances are slimmer:
459=9-9, 464=15-15-17-17, 413b=24-24.

This combination would probably not consist of recLOHs in a P312+ person, as the values are simply too common:
459=9-9, 464=15-15-17-17, 413b=23-23.          <----- Maybe 459=9-9 is a recLOH, but it could just as well be a normal one-step mutation from 459=9-10.

This combination has two uncommon double values for R1b1a2 and would likely be two recLOHs, UNLESS his close matches had one of these two doubles already:
395s1=16-16, 413b=24-24
In other words, if the person's close matches had 395s1=16-16, 413b=23-24, then you would conclude a normal step mutation for 413 rather than a recLOH for both markers.
On the other hand, if the person's close matches or closely related haplogroup had 395s1=15-16 and 413b=23-24, then you could conclude that a single recLOH event changed both markers in one fell swoop, at some time in the past in this man's particular lineage, from 395s1=15-16 to 16-16 and from 413=23-24 to 24-24.

This combination has unusual values, but only has one uncommon double value for R1b1a2 and is indiscernible from a normal step-up mutation in 395s1:
385=11-13, 459=8-10, 464=13-15-16-17, YCAII=19-24, CDY=34-39, 395s1=16-16, 413=23-25          <----- We have no easy way of knowing if 395s1 is a recLOH.










What is Panel 5 Palindromic Pack?

This is an approximately $100 test.  If you want to order this, you can't do so from FTDNA's website.  Instead, you need to call FTDNA customer service, 713-868-1438.  The test results are byzantine and offer no interpretation.  All you will have are a list of STR values for some palindromic markers.  I'm still puzzling over the Panel 5 results for kit 233764 in haplogroup D.  (I ordered the Panel 5 test for him purely out of scientific curiosity, as RecLOHs are fascinating to me.)  You can see his rather cryptic results here:  http://forums.familytreedna.com/showthread.php?t=31776&highlight=RecLOH
The question was, did a recLOH event cause 233764's doubled markers?  The answer still eludes me, as the panel 5 results had some doubles but other non-doubles.  Possibly some of the markers had recLOHs in the distant past, but one or both copies have since mutated downwards or upwards.










What are the other palindromic markers, not included in the standard 67-marker test, that Panel 5 Palindromic Pack looks into?

Here is the full list based on Thomas Krahn's diagram, © 2006 Thomas Krahn:

On Palindrome 1:
DYF397, DYF401, DYS459, DYF385, DYS724 also known as CDY, DYF371X (one half of 371), DYF408, DYF399X, DYS464X (one half of 464), and DYS725

On palindrome 2:
DYS464X (the other half of 464) and DYS725

On palindrome 3:
DYF397

On palindrome 4:
DYF411 and DYS385 (not to be confused with "DYF385" on P1)

On palindrome 5:
YCAII, DYF395s1, DYF371X (the other half of 371 that includes DYS425), and DYF408

On palindrome 8:
DYS413










What is special about marker 389?

The DYS389 marker is the most confusing multicopy marker.  Different companies and labs have different names and different ways of reporting the readings for this marker, adding to the confusion.  This marker is not involved in recLOHs but is often presented in dash format (for example, 13-29) like the palindromic markers are.

This marker does not consist of two copies (two amplified locations on the Y chromosome), as other multicopy markers do.  Instead, there is one copy (one amplified location) that is read differently in each direction (forward or reverse), resulting in one allele (one repeat value) or two alleles (two repeat values), depending on which direction the marker is read.

Primer L, the forward primer for 389, binds at one specific location on the Y chromosome.
Primer R, the reverse primer for 389, binds at two specific locations.
This results in two readings:  One for the shorter 389i or 389I or 389-1 fragment, and another for the longer 389ii or 389II or 389-2 fragment.
The shorter fragment is a subset of the longer fragment.

389ii refers to the total length of 389. A mutation in the smaller 389i fragment that is a component of the larger 389ii fragment will necessarily affect the reading for the larger fragment as well.  In other words, when there is a one-step mutation in 389i, there will be a corresponding one-step mutation in 389ii.  However, the reverse is not true.

One way to state the allele mutations separately is to measure the change between 389i and 389ii (that is, 389ii minus 389i) as a separate marker.  The name for this measurement is 389b or 389ii-i or 389delta.  A mutation in 389b - the part of 389ii that is beyond the 389i region - changes the repeat value of the entire marker, without affecting the repeat value of the smaller 389i component.


Here is an analogy:

You have a train with 29 railcars.  The first 13 railcars are red.  The other 16 railcars are blue.  Assume that red must stay with red, and blue with blue.
389i = the number of red railcars.
389ii = the total number of railcars, regardless of color.
Initial reading is 389i=13, 389ii=29.
If you insert a red railcar in the 14th repeat position, then the readings will change to 389i=14 and 389ii=30.
If you now insert a blue railcar at the 31st repeat position, then 389i will remain the same, at 389i=14, whereas 389ii will change to 31.
If you now take out a red railcar, 389i will drop to 13 and 389ii will drop to 30.


If you see a kit in which 389ii apparently mutated from 29 to 31 sometime in the past, but you don't look at 389i or 389b, then you have no way of knowing if:

389i increased two steps (one two-step mutation),
OR
389i stayed the same and the region beyond 389i increased two steps (an entirely different two-step mutation),
OR
389i increased one step and the region beyond 389i increased one step (two one-step mutations).


See more DYS389 details here:
http://www.smgf.org/pages/dys389.jspx
http://www.genebase.com/learning/article/45
http://www.genebase.com/in/dnaLearningCenter.php?id=19


Sometimes, a genetic cluster will tend to have a mutation in 389i but not in 389b; for example, those with signature 49-2329222-1112, who tend to have 389i=12 and normal 389b values, with 389ii typically 28.

Sometimes, a genetic cluster will tend to have a mutation in 389b but not in 389i; for example, those with signature 21-5909-A, who tend to have 389b=17, with 389ii typically 30.  Searching for their off-modal value of 389b=17 also involves looking for occasional instances of 389=12-29 or 389=14-31.

Sometimes, a genetic cluster will tend to have mutations in both 389i and 389b; for example, those with signature z1014-B, who tend to have 389i=14 and 389b=15, with 389ii typically 29.