This Blog is Part 3 documenting my learning process of phasing my DNA raw data using:
- My raw data results, along with my mother’s and 2 sisters’ raw DNA data
- MS Access to speed up the process
- A Whit Athey Paper on Phasing when the data from one parent is missing
- The work of M MacNeill on my raw data results.
Part 1 and 2 Recap
- I imported 4 sets of raw data into Access from AncestryDNA after taking out the zeros that the Excel software produced for the no-calls.
- I used Access Queries to apply 3 Whit Athey Principles. This resulted in many phased bases for me and my 2 sisters.
- I put the phased A’s, G’s, C’s and T’s for each siblings into 2 new columns for each sibling
- This resulted in 6 new columns. The first 3 of these six were for the paternally based bases. These resulted in a pattern which was either in the form of AAB, ABA, or ABB.
- The Athey Paper did not emphasize the AAA pattern or considered it a non-pattern. While specific AAA results within another pattern area are by chance, there are other areas where 3 siblings match the same grandparent where there will be an AAA-only Pattern.
- I separated my results into 3 patterns using Access: AAB, ABA, and ABB
- For each of those results, I noted where those patterns changed. I did this by looking at the ID numbers. Breaks in the ID numbers were considered changes.
- However, there were some cases where the changes occurred around missing bases. For these, I went back and noted a more precise position of the pattern change based on where the change would be if the missing base were to be filled in.
- I Made a preliminary bar graph using the first 3 paternal changes. These crossovers were mapped to myself and 2 sisters.
- Using the 3 patterns I developed Access queries to fill in the missing bases in the 3 paternal pattern areas.
So those were the 10 easy steps. Actually step 10 was difficult as there was quite a bit of refining the Access queries and quality checking the results. I needed 2 queries for each of the pattern areas. However, once I had the queries, it was the push of a button to update missing parental-received bases for 3 siblings within over 700,000 lines of DNA.
Back to Athey
This portion of the Athey Paper appears to apply to where I am now:
For some of the unfilled cells on the mother’s side of the table, we can fill in the alternative (other) base from the corresponding location on the father’s side of the table. That is, we know that the sibling with an empty cell got one base from the father, but the alternative base from the mother. Therefore, after the use of the Dad pattern fills in more cells, a newly filled – in cell in the father’s side of the table gives rise to a filled – in cell in the same position on the mother’s side–the alternative base to what was on the father’s side.
Unfortunately, I’m not sure what is meant above. My guess is that this relates to Principle 3:
Principle 3 — A final phasing principle is almost trivial, but it is normally not useful because there is usually no way to satisfy its conditions: If a child is heterozygous at a particular SNP, and if it is possible to determine which parent contributed one of the bases, then the other parent necessarily contributed the other (or alternate) base. This principle will be very useful in the present approach.
So now that missing paternal bases have been determined based on the patterns, it should be possible to fill in missing maternal bases for heterozygous children. First, I’ll do a Query to see if I can locate this situation. I’ll take my most recently updated Dad ABB Pattern Table update and query that. I’ll look at the situation where there are heterozygous results. Then, I’ll look at spots where there are missing bases from Mom.
Fortunately, I was able to come up with a slick looking Query for this situation:
Plus the Query design has some nice symmetry. The first criteria row of the query is for my (Joel) DNA. Reading across, it says Joel is heterozygous because my allele 1 does not equal my allele 2. Then it says that I have a base from Dad but not from Mom. This will show areas where the mom bases are missing in this heterozygous child situation.
The truncated fields above are Joel Allele 1, Joel Allele 2, Sharon allele 1&2, Heidi allele 1&2. The next 3 columns are Joel, Sharon and Heidi from Dad. Then Joel, Sharon and Heidi from Mom (the last 3 columns). This shows that there are almost 12,000 of these Mom bases to fill in. Above the blue line are Heidi’s bases missing from Mom. Heidi is TC (heterozygous) on that line. Her Dad base is T. I love these binary problems. They seem well suited for the computer. That means that a query could not be too difficult to update almost 12,000 records. So Heidi’s Mom base will be C above the blue line. At the blue highlighted area, I am TC and my Dad base is C. My Mom base will be T on the blue line.
Looking for a Good Query to Fill In Mom Bases from Dad Bases
First, I copied my ABB Table to a new Table called tbleMomBaseFromDadBase. I will want to update that table with a new Update Query. I already have the first part of the query. Now I need my thinking cap. Even better than thinking, I can look at what I did before. Here is my old query.
This is difficult to see, but I split the problem into 2 alleles. What this says is when Sharon has a base from her mom and Sharon’s allele 1 is not the same as the base from her Mom, pop that allele 1 into her base from Dad slot.
For our situation we are doing the opposite. So we will switch Mom and Dad. This time we are using our Dad results to get some Mom results. I’ll also add a criteria to make sure the Mom result is Null, so I’m not overwriting anything. It will just be an extra precaution.
Basically, I want to make sure Heidi has a base from Dad and not from Mom. In that case, when her allele1 is not equal to her base from Dad, put that allele 1 in as her base from Mom. Drawing upon my vast experience in this area of about 1 week, I get this:
When I preview the results, I get about 6,000 lines which is half of my previous query, so that seems OK. I’ll go ahead and update my new Table. I renamed my Query to qryMomBaseFromDadBaseAllele1 and copied it to do the same thing with Allele2. I’ll change the Allele’s 1’s to Allele’s 2 in the Query design. First I’ll do a Select (non-updating) Query to show what I’ll be updating with the allele’s 2.
Here I added the ID numbers, so I can make sure my update went well.
Here is my Allele2 Update Query with the 3 siblings included:
The results:
In the far right column is the Base Heidi got from Mom. It was updated on lines 2292, 2295 and 2299. In each case Heidi’s Paternal Base was T and the Maternally derived Base from Dad was C.
Here is my corresponding filled in Mom Base:
My Dad’s T’s in 6 columns from the right were used to fill in the missing C’s in 3 columns from the right. Doesn’t it seem a bit ironic? Even though my dad was not tested for DNA, his “results” from this process are used to find the DNA I got from my mom who was tested.
A Premature End to This Blog and a New Beginning
This will be one of my shortest Blogs. I was both awaiting and not awaiting my brother’s DNA test results. Those results came in this week. The reason I was not awaiting was that I knew that I would need to re-start the raw data DNA phasing process once his results came in. With that, I’ll end this Blog and start a new one.