Introducing the Borland Genetics Segment Lab

 

It occurred to me today that I have not yet done a tutorial on how to use the Borland Genetics “Segment Lab” tool.  Introduced as part of last October’s anniversary “Fall Features Release,” it’s one of the newest tools on the site, and probably also one of the most under-utilized.  Hopefully, this demonstration will show how it’s also one of the most valuable tools on the site.

The Segment Lab is Borland Genetics’ native chromosome map/phase map editor.  While a far cry from a robust chromosome mapping solution like DNAPainter, it has specialized features that I designed to make our task of ancestor DNA reconstruction easier, and to help maximize the reconstruction coverage of our output ancestor kits.

Before we dive into using the tool, I want to point out that the Segment Lab is NOT a tool that creates a phase map from scratch.  At Borland Genetics, that’s what the HIR Mapper does, and when it comes to mapping matches from outside the Borland Genetics platform, DNAPainter is the go-to website to accomplish this.  Borland Genetics is compatible with DNAPainter and DNAPainter maps work great with the Extract Segments tool.

The next preliminary matter is understanding the difference between a chromosome map and a phase map.  All phase maps are chromosome maps, but the opposite is not true.  A phase map, per Borland Genetics terminology, is a specific kind of chromosome map, where the chromosomes of a phased DNA kit are mapped according to the parent from which each block of DNA was inherited, i.e. the block’s parental phase.  Within each block, there are segments of DNA that were inherited from more distant ancestors on the same side from which the block was inherited.  Shared segments with cousin matches are typically mapped on a more broader chromosome map, where phase maps only concern themselves with whether entire blocks of DNA are paternal or maternal to the subject of the map.  Most Borland Genetics workflows incorporate the use of phase maps, for two reasons:  1) The Borland Genetics reconstruction method requires phasing prior to mapping; and 2) In DNA reconstruction, it is best to step back one generation at a time so that decisions can effectively be made when determining how many generations back each recombination point that divides our segments occurred.  That’s a lot to digest, so let’s get some hands-on experience working with phase maps in Borland Genetics.

One Borland Genetics workflow in which phase map editing is critical is the “Reverse Phase” workflow, where the blocks of DNA inherited by a subject donor’s child and that child’s hypothetical evil twin are compared side-by-side to determine which blocks were inherited from each of the donor’s parents.  Since the in-phase child and the evil twin inherited opposite blocks of chromosome from the subject donor, the resulting phase map will have a signature checker-board pattern where recombination points in the most recent generation mark visible “switches in streams” of the subject donor’s paternal and maternal DNA.  For convenience/consistency, phase maps generated by the Borland Genetics “HIR Mapper” tool are color coded as blue or pink based on whether blocks correspond to paternal or maternal inheritance.

The image below is a portion of an automatically generated phase map created by the Borland Genetics “Creeper” representing the parental phase of the blocks of DNA that my cousin Penny passed to her child April, and would have passed to April’s hypothetical evil twin.  By definition, all of this DNA is maternal relative to April (or to her evil twin), as the DNA is coming from April’s mother Penny.  The parental phase we are concerned about when mapping is relative to Penny.

The top bar of each chromosome represents the maternally phased DNA of April, whereas the bottom bar represents the DNA that Penny did NOT pass to April.  Where blue blocks appear on the phase map corresponding to Penny’s paternal copy of a block of chromosome, on the bottom row of each chromosome, pink blocks appear where Penny passed April’s evil twin inherits DNA from the opposite copy of Penny’s chromosome (in this case maternal).  Where April zigs, her evil twin zags, so to speak.  If April inherited DNA from her maternal grandmother (Penny’s mother), then April’s evil twin necessarily inherited DNA from her maternal grandfather (Penny’s father).

Black bars indicate that phase cannot be determined due to a lack of matches in the Borland Genetics database.  There are also some green bars where there are matches in the database but their family side relative to Penny is unknown to the “Creeper” that auto-generated the map.  Some maps will also have red bars where a span of chromosome appears to have conflicting phase information (usually due to pedigree collapse, endogamy, or a failure of the Creeper to recognize a recombination point that divides blocks of opposite phase).

I found this phase map in my Phase Map Locker.  It appears in three phase map lockers, actually:  1) the overall Phase Map Locker accessible from the Tools tab; 2) the Phase Map Locker attached to the evil-twin kit linked to Penny’s profile; and 3) the Phase Map in the Reverse Phase Smart Project in which it was generated.  If the phase map is associated with a Smart Project, it is usually easiest to access phase maps from the project phase map locker (less maps to comb through, less clicks to get there).

After selecting a phase map that we want to work with, we first select the “Graphical Preview” button beside the description of the map.  We are then taken to a screen which displays some basic information about the map (a link to the kit the map corresponds to, a link to the subject donor, a link to the project, and a graphical representation of the phase map with the above-described color scheme).

At the bottom of this screen, there are a few green navigation buttons to choose from.  Borland Genetics subscribers will see a top navigation button that says “Launch Segment Lab.”  Our phase map editing adventure begins here!

When you first enter the Segment Lab, you are transported to the first segment of your phase map (in this case it is a segment on April's chromosome 1 that the Creeper identified as maternal relative to Penny (inherited from April's maternal grandfather Pete).  The first two sections of the Segment Lab screen, pictured below, provide some general information about the segment being explored.

The first piece of information displayed is the segment type, indicating the source of phase map where the segment is found.  In this case, the map was automatically generated by the Creeper from a Reverse Phase Smart Project.  Other possible sources are ordinary phase maps generated by HIR Mapper tool (whether manually or as part of a Smart Project), or phase maps generated by DNAPainter or manually in Excel using the DNAPainter template.  The next lines display the name of the map (with a hyperlink back to the graphical preview screen), and then information about the donor and kit to which the segment/map pertain.  Also, the fact that the map was generated in Build 37 coordinates is indicated.

The second section provides the chromosome number, and the start and stop positions that define the segment.  The approximate statistical length of the segment in cM is also provided, calculated using the Borland Genetics "Crude Mary" polynomial Marey map.

Scrolling down to the next section of the screen (pictured above), a list of matches to this segment in the Borland Genetics database is provided.  Since this is a segment from a special Reverse Phase map, for convenience, the matches on the opposing evil twin segment are also displayed below the matches to this segment.  That way, errors or conflicts in the phase assignment can be easily flagged and fixed.  Also, when the Creeper could not automatically detect whether the segment is paternal or maternal because matches on the segment are too distant, sometimes you will recognize a distant cousin match and be able to manually assign phase accordingly.  Also, please note that start and stop positions where each match on that segment begins and ends is provided.  How to use that information will be discussed later, but at this point, just be advised that the match may or may not extend beyond the boundaries of that segment but that the start and stop positions on the chart will be bound to the dimensions of the segment in question.

The next section of the screen (applies only to phase maps generated using the HIR Mapper or other Borland Genetics tools) displays the segment's horizontal parity.  That is, it tells whether the segment to the left and to the right are paternal or maternal, if known.  Here, the segment is on the far west tip of the chromosome so there is no segment before it, but the neighboring segment to the east is paternal.

Vertical parity information, i.e. the phase of the opposing segment on the opposite evil twin segment, is not displayed in this section, but users can easily scroll down to the graphical display of the phase map at the bottom of the screen to make that determination if necessary.

This is not a particularly interesting segment to me since I can tell from the names of the matches across the segment that the Creeper correctly identified the segment as maternal.  So let's take a look at a more interesting segment, the black (empty) one I circled with a highlighter here on chromosome 4:

I'm going to scroll down to the bottom of my screen and click on that segment I want to explore on the navigable phase map.  Note, sometimes this will kick you out of the Borland Genetics site and you will have to log in.  However, this will only happen once in a session, so once you log back in, the problem will not persist.  (I'm hoping someone more experienced in JavaScript programming can help me fix that glitch sometime in the near future.)

It is not obvious that the map does anything when you click it, but clicking that black segment will in fact transport you to the segment.  Why have I chosen this segment to explore?  First, because it is sandwiched between two segments that were already determined by the Creeper to be paternal.  My inquiry is whether it is OK to assume that this span of chromosome is also paternal like his neighbors, or whether it's possible that there were two recombination points in the most recent generation and that the segment might actually be maternal.

When I scroll down to the horizontal parity information for this segment, I am shown some additional information that is very useful:

The system calculates on the fly, based on the statistical length of the separation of the nearest same-side segments, the odds that this sandwiched segment is also the same parental phase as its neighbors.  In this case, the system indicates that the odds of this span of chromosome being paternal is 95.7%.  The tool also recommends that I only rely on this statistic where the likelihood is greater than 95%.  I will therefore flip the segment to paternal on the next section of the screen.

From the drop-down, I change "Unknown" to "Paternal" and then I simply click the "Modify Phase" button, which will make an update to the phase map stored in the Phase Map Locker database.  By default, the entire selected segment will be changed to paternal, but if that is not the desired outcome, you can adjust the trim points to just a portion of the segment.  The trim point drop-downs are pre-populated with the start and stop positions of any matches to the segment (or to the opposing evil twin segment), although since we are 95.7% certain that the entire gap can be flipped to paternal, we will ignore the option to set trim points for now.

After clicking "Modify Phase," and then waiting for the page to refresh, I then scrolled down to the navigable graphical preview of the map at the bottom of the page and now it looks like this:

Notice that the opposing evil twin segment was not automatically flipped to maternal.  We should also do that at this point, in the same manner we modified the in-phase segment (by clicking on the gap between the pink segments and then changing the phase designation to maternal).  Please note that it is not recommend that you bypass reviewing the statistical prediction and just eyeballing whether to flip a gap to the phase of its neighbors.  This is because the contour of each chromosome is not uniform and what may look like a small gap on one part of a chromosome, may actually be large in terms of statistical probability for recombination.  This is all taken into account by the tool so you don't have to guess.  Also, please do not rely on a reduced percentage of 75%, however tempting that may be.  If you do, then 1 OUT OF 4 TIMES YOUR GUESS WILL BE WRONG!

As our next case study, let's look at this segment on chromosome 2:

It is one of a few segments that are sandwiched between maternal segments on an evil twin chromosome map.  I'm choosing this segment to illustrate a few different points.

First, note that in the matches section, the matches to the evil twin segment appear on top, and the matches on the opposing in-phase kit are listed below.  This is the case so as to constantly remind the user which segment we are pondering.  The matches to the focus statement will always be listed first.

At this point, you are probably saying, "Wait just a minute.  I thought you said this was an empty segment!"  Well, it was when I created the map in 2020, but there have been a lot of new kits uploaded to the database since then.  You will see that the in-phase kit matches April, but that is trivial because the kit was created by phasing April against Penny and we are only concerned with matches that are related paternally or maternally to Penny, and not whether the segment is a match to Penny's daughter.  But we see that kits for a paternal relative of Penny that I recognize (Burnette) and for a maternal relative (Naida) have been uploaded since I first made this map in 2020.  We'll get back to them, but first, please take note that the prediction for designating the entire span of chromosome to maternal like its neighbors is reported as 51.9% which is basically like flipping a coin.  We will not be doing that.  Instead, notice that on the evil twin segment, the match to Naida runs to about 141 MBP whereas the match to Burnette begins at about 141 MBP.  The opposite is true for the in-phase segment matches.  What does this mean?  We have discovered a recombination point, where the phase changes within the span we are exploring.  This is where selecting trim points comes into play.  In the phase designation portion of the tool below, we will modify the phase to maternal, but only on the west portion of the span that now matches Naida.  We do that by adjusting Trim Point 2 to the end position of the match to Naida.  If you do any video editing on your cell-phone, think of the trim points like where you want to crop your video to get rid of footage and the beginning and end that show you setting up your shot or turning off the camera!  I show my trim point settings on the image below:

One final note, and then I'm going to go on and examine every segment on this map that hasn't yet been assessed.  This map was created automatically via a single-child Reverse Phase workflow.  Since there was only one child involved, recombination points could only be automatically discovered where matches to kits in the database begin and end, and so where there are no matches across a span of chromosome, there may be hidden recombination points like the one we discovered.  However, had it been a multiple-child Reverse Phase project or a Visual Phasing project from which the Creeper generated the map, these hidden recombination points are not likely to occur on a map ONCE THE SEGMENT EXPANDER TOOL HAS MADE ITS AUTOMATED EDITS.   The Borland Genetics Segment Expander uses sibling relationships to automatically create invisible segment breaks in the phase map at auto-detected recombination points.  Even when only two siblings are involved, the Segment Expander will discover approximately 99% of such recombination points.  It is almost always safe in multiple-sibling workflows therefore to rely on a segment being a single segment and you can just ignore the trim points with the exception of the 1% of the time when you might notice a conflict.

I hope this is helpful to everyone who uses my tools.  There are a lot of other nuances to mapping and reconstruction generally that I learned through almost a decade of experience doing this.  If you have any questions, don't hesitate to ask me.  I respond best to posts made in the Borland Genetics Users Group on Facebook.