Trait Predictor 16_5 is out!

[AndreiDNA]

v16_5 version of my Trait Predictor comes out today!
Buy it here:
https://andreikh.itch.io/trait-predictor
What I added (after the 16_4 version): Goucher syndrome, Congenital adrenal hyperplasia , Wilson syndrome, Psoriasis polygenic score, Obsessive compulsive disorder polygenic score, added more SNPs for leukemia polygenic score

I haven't updated my Trait Predictor in 8 months because I thought there is nothing more to add. I had to think long and hard to find conditions I still haven't sufficiently covered. The polygenic score for OCD was VERY experimental. The genes I selected for it totally make sense as far as causing OCD but there just isn't enough research into this condition.
If I had to name the most robust and reliable conditions that my Trait Predictor predicts, it'd be hemoglobin E disease > Multiple sclerosis > Alzheimer's > everything else. Just because these conditions have so much research behind them and there was a lot for me to go off there. The Trait Predictor condition with the most SNPs in the polygenic score is bipolar 1. But a lot of these snps are experimental, discovered early and without replication studies. I tried to incorporate both recently discovered, robustly correlated (such as ADCY2 or TRANK1) and experimental variants (Like DRD2 and DISC1) into my polygenic score. Unfortunately a lot of these conditions are very poorly researched, which is the problem. For OCD I had basically nothing to work with so I went hardcore experimental and used gene function as my compass.

You can check out different reports with trait predictor from different time periods on my trait predictor website (at one point my trait predictor was a webapp and you could purchase it there. I have since repurposed it as an executable and took down the webapp)
https://andreikharchenko1.github.io/Trait-Predictor/
You can check out the different archaic reports that my Trait predictor generated. There's even a neanderthal result on the site.
If you're having trouble running the 16_5 version of trait predictor (could be due to an antivirus or weak processor), there's a simple fix
just run "
!chmod +x AndreiDNA_TraitPredictor_v_16_5_Linux


!./AndreiDNA_TraitPredictor_v_16_5_Linux"
in google colab and it will work. To state the obvious you first need to upload the linux executable and your DNA file in the workflow. In case you can't figure out how to do that I'll just leave the 16_4 version on itch as well, since I know it runs without problems.

I am raising the price for Trait Predictor executable to 30$. If you had purchased it previously you can get the update at any time on itch. If you still haven't purchased it and don't want to spend money on it, you can purchase a report for 4$ here:
https://docs.google.com/forms/d/e/1FA...

Sample report of an Anonymous user (He's a White american with some mexican ancestry), AncestryDNA file. If your sample is Ancestry your report will likely look similar (will contain the same SNPs, genotypes will obviously be different)
https://andreikharchenko1.github.io/AnonAmerican/
^^^^
My trait predictor is optimized for Myheritage, 23andme, and ancestrydna raw files. If you tested with these companies most of the SNPs you need for analysis will be in your sample.

As I'm typing this I noticed the date on my current version says its from 2/31/2026 instead of 2/21/2026. Minor bug that might be funny to some. I don't feel like spending time fixing it now it doesn't really hurt the report.

[Beowulf]

Should i send you an email if i'd like mine to be updated?

[AndreiDNA]

Should i send you an email if i'd like mine to be updated?

I think i got your email. I had someone message me 2 hours ago

[Beowulf]

I think i got your email. I had someone message me 2 hours ago

Oh! I haven't sent any email myself yet lol. However, i assume i should send you one then?

[AndreiDNA]

Oh! I haven't sent any email myself yet lol. However, i assume i should send you one then?

yes go ahead

[cass]

In the phenotypic description itself, there probably aren’t any differences.

Is the Y chromosome determination detailed, or is it at a general level?
This individual's Ydna is : --> F --> K --> P --> R --> R1a --> Z645

I had the impression, based on other predictors, that it was at a much lower (more specific) level, namely R1a-M458.

[AndreiDNA]

In the phenotypic description itself, there probably aren’t any differences.

Is the Y chromosome determination detailed, or is it at a general level?
This individual's Ydna is : --> F --> K --> P --> R --> R1a --> Z645

I had the impression, based on other predictors, that it was at a much lower (more specific) level, namely R1a-M458.

the Ydna thing is not precise at all. You don't need my trait predictor for that, there's already yseq cladefinder. Anyway im not sure how i feel about selling the executable at all anymore. I think i should make it reports only. Im working on coding the webapp for automated reports right now

[AndreiDNA]

I am changing my policy toward my Trait predictor.
This project is the best thing I have made in my life. It has 78 thousand lines of code. Most of them were generated automatically but I wrote at least 10k myself. It was a collosal piece of work. I started working on it in 2023 and made the last update to it yesterday.
I've been selling the executable for 25$. This is ridiculous. This is easily the best tool for health and traits that exists and I'm selling THE EXECUTABLE for the price of a single myheritage trait report with 20 snps. Since trait predictor's inception it's been sold 252 times. I knew full well that I don't have the reputation and aura to sell a tool like that for what it's actually worth so I sold it for 25$.
I had to file taxes this week and it hit me. My business model is not sustainable. I can't keep doing this forever. I'm not making ends meet. People have been asking me, what is my job. This is my job. This is what I do and think about 24/7. I always wanted this. The reason i went to college for CS in the first place is to do this. I graduated last year and have been working full time on my tools. Since this is what I'm doing I want to earn a living. It's just math. If you sell something worth 300$+ for 25 bucks you're eventually going to run out of customers.

So now, although it is way too late, I am changing my policy.
From now on Trait predictor executable will be worth 400$.
I'm not making it unafforable. This is how much an executable should cost.
A single report will be worth 10.5$ on my new website i've been setting up today.

I worked hard on the site. I incorporated an SQL database to store info on users and their results. I incorporated threads meaning you can close the page/ log off/ refresh/ whatever and your file will still be processing. Two or more people can initiate processes simultaneously. Since I'm using pythonanywhere to host and I can't afford their membership, i'm limited by 500Mb of server space. Don't worry, this won't restrict you - DNA files are deleted from the server as soon as the process is over and only the .html reports remain forever. They weigh very little but if - say - 2000 people will buy this service, I will eventually have to delete some results. Don't worry - there's an option to download them on the dashboard. If you're afraid of this scenario you can just download your results and save them somewhere.

On the trait predictor report page I included a demo account with login: test2 pass: test2 which you can use to test out the webapp.

On my computer, it typically takes about 10 minutes to process a 50Mb file with trait predictor. But with pythonanywhere, it will take longer. I did an experiment with a 50Mb file and it took it nearly an hour to finish processing it. If multiple people upload files at the same time it will take even longer. But once again, you can just close the site and come back later - the results will be there.
https://traitpredictor.pythonanywhere.com/login
^this is the site

Also I suggest to everyone to download their results as soon as possible, as there might be vandalists writing scripts to auto-create accounts and inflate the SQL database to the point it occupies the whole 500Mb. That's an unlikely scenario but you have to consider every possibility.

Final statement - I won't be doing manual trait predictor reports anymore. It's automated now. I'll be removing the manual option from my google forms

[George Weah]

...

You need to reach clients who'd be willing to pay more. Collaborate with European-based companies like GenePlaza, Illustrative etc or medical research labs anywhere. Professionals.
Your average genealogy enthusiasts (like me and the other members) won't make you rich (my two cents, no pun intended). Unless you have a huge client base, which is going to take time and lots of marketing.

Talent never gets rewarded, not proportionately in any case. Marketing does. (In your case , leveraging your knowledge requires a more business-like client base).