John Belmont
@jwbelmon.bsky.social
#MD, #PhD | #Pediatrics | #Medical #Genetics | #Human #Genetics | #Cardiovascular | #Congenital | #Genome #Sequencing | #HealthEcon | #Causal #Inference
belmontgenetics.org
belmontgenetics.org
Pink Brugmansia - they love the warm damp mornings 🌱
September 22, 2025 at 12:24 PM
Pink Brugmansia - they love the warm damp mornings 🌱
Pink Brugmansia - they love the warm damp mornings 🌱
September 22, 2025 at 12:22 PM
Pink Brugmansia - they love the warm damp mornings 🌱
Ai Weiwei catches the moment long before the rest of us feel it
August 24, 2025 at 1:34 PM
Ai Weiwei catches the moment long before the rest of us feel it
Thinking about genetic effects in terms of causal DAGs leads to ways of dealing with confounding that each have complementary strength. Matching, inverse probability weighting, regression, and doubly robust regression all give comparable estimates of effect
August 19, 2025 at 2:37 PM
Thinking about genetic effects in terms of causal DAGs leads to ways of dealing with confounding that each have complementary strength. Matching, inverse probability weighting, regression, and doubly robust regression all give comparable estimates of effect
Power is pretty comparable to 'forward' association tests. But under certain circumstance can exceed typical tests because it is exploiting both differences in allele frequency as well as the collider induced signal
August 19, 2025 at 2:37 PM
Power is pretty comparable to 'forward' association tests. But under certain circumstance can exceed typical tests because it is exploiting both differences in allele frequency as well as the collider induced signal
Here is the key idea. If you have a known cause of a phenotype, like a PGS, then it's possible to leverage that to infer whether another factor, like rare variants in a candidate gene, are also causes. A statistical but non-causal association emerges when you condition on the phenotype.
August 19, 2025 at 2:37 PM
Here is the key idea. If you have a known cause of a phenotype, like a PGS, then it's possible to leverage that to infer whether another factor, like rare variants in a candidate gene, are also causes. A statistical but non-causal association emerges when you condition on the phenotype.
Five more today. It’s warm and muggy in the early morning.
July 13, 2025 at 12:02 PM
Five more today. It’s warm and muggy in the early morning.
A day that begins with a night bloomer is promising 🌱
July 12, 2025 at 11:59 AM
A day that begins with a night bloomer is promising 🌱
Or do you like this one
May 25, 2025 at 11:26 PM
Or do you like this one
These were just right for picking. Guests at the garden this evening. 🌱 which picture do you prefer? Left or right?
May 25, 2025 at 11:21 PM
These were just right for picking. Guests at the garden this evening. 🌱 which picture do you prefer? Left or right?
I wasn’t sure if cereus would bloom on an apartment balcony. Five flowers this morning. Yay! 🌱
May 16, 2025 at 9:23 PM
I wasn’t sure if cereus would bloom on an apartment balcony. Five flowers this morning. Yay! 🌱
Loaded with green tomatoes
May 5, 2025 at 9:09 PM
Loaded with green tomatoes
Smithers Park in Houston
May 5, 2025 at 2:09 AM
Smithers Park in Houston
Twenty tomato plants now big enough to support. Beginning to flower.🌱
April 12, 2025 at 5:12 PM
Twenty tomato plants now big enough to support. Beginning to flower.🌱
So here we are on Sunday evening
April 6, 2025 at 11:55 PM
So here we are on Sunday evening
hr@opm.gov is the actual email address they are using and maybe you would like to tell them what you did this week, too
February 23, 2025 at 7:44 PM
hr@opm.gov is the actual email address they are using and maybe you would like to tell them what you did this week, too
Walking the dog this morning and this caught my eye
February 16, 2025 at 2:21 PM
Walking the dog this morning and this caught my eye
Yesterday I saw a coachwhip snake snooping around in the barn, so I know Spring is here.
February 9, 2025 at 4:14 PM
Yesterday I saw a coachwhip snake snooping around in the barn, so I know Spring is here.
Two boxes weeded and tilled. Ready for tomatoes 🍅. Seedlings looking good. Room for Spring garden optimism.
February 8, 2025 at 10:01 PM
Two boxes weeded and tilled. Ready for tomatoes 🍅. Seedlings looking good. Room for Spring garden optimism.
Let:
𝑌 be the door with the prize
𝑋 be the door chosen by the contestant
𝑍 be the door opened by the host
When the game starts X⊥Y, and Z is influenced by both X and Y; but after Monty selects Z the DAG can be updated
𝑌 be the door with the prize
𝑋 be the door chosen by the contestant
𝑍 be the door opened by the host
When the game starts X⊥Y, and Z is influenced by both X and Y; but after Monty selects Z the DAG can be updated
February 5, 2025 at 5:15 PM
Let:
𝑌 be the door with the prize
𝑋 be the door chosen by the contestant
𝑍 be the door opened by the host
When the game starts X⊥Y, and Z is influenced by both X and Y; but after Monty selects Z the DAG can be updated
𝑌 be the door with the prize
𝑋 be the door chosen by the contestant
𝑍 be the door opened by the host
When the game starts X⊥Y, and Z is influenced by both X and Y; but after Monty selects Z the DAG can be updated
Following the temporal order: the contestant picks a door, presumably independent of where the prize is. Then Monty opens one of the two other doors which always has a goat. Monty offers the contestant the opportunity to switch doors. Should they switch?
Here's the DAG
Here's the DAG
February 5, 2025 at 5:15 PM
Following the temporal order: the contestant picks a door, presumably independent of where the prize is. Then Monty opens one of the two other doors which always has a goat. Monty offers the contestant the opportunity to switch doors. Should they switch?
Here's the DAG
Here's the DAG
The Monty Hall Problem is a classic.
Thank you, Chad Shaw (Rice U STAT530), for pointing out that it can be viewed through the lens of graphical models. I'm posting it here because it's fun (and selfishly to reinforce the factorization for me).
There are 3 doors: one prize and two goats
Thank you, Chad Shaw (Rice U STAT530), for pointing out that it can be viewed through the lens of graphical models. I'm posting it here because it's fun (and selfishly to reinforce the factorization for me).
There are 3 doors: one prize and two goats
February 5, 2025 at 5:15 PM
The Monty Hall Problem is a classic.
Thank you, Chad Shaw (Rice U STAT530), for pointing out that it can be viewed through the lens of graphical models. I'm posting it here because it's fun (and selfishly to reinforce the factorization for me).
There are 3 doors: one prize and two goats
Thank you, Chad Shaw (Rice U STAT530), for pointing out that it can be viewed through the lens of graphical models. I'm posting it here because it's fun (and selfishly to reinforce the factorization for me).
There are 3 doors: one prize and two goats