Amy Inkster
@amy-inkster.bsky.social
Postdoc in the Cardenas lab @Stanford Epidemiology, PhD in Medical Genetics @UBC & @BCCHResearch in @robinson_lab. Epigenetics and early development.
Congratulations Sam!! This just popped into my feed. So happy for you!! 💕
September 5, 2025 at 4:43 PM
Congratulations Sam!! This just popped into my feed. So happy for you!! 💕
If this piques your interest, check it out! Thank you to my co-authors Hannah and Wendy, and to all members of the Robinson lab whose work contributed to these lessons and who engaged in many extremely detailed lab meeting conversations about everything from normalization methods to plot colours.
April 24, 2025 at 8:56 PM
If this piques your interest, check it out! Thank you to my co-authors Hannah and Wendy, and to all members of the Robinson lab whose work contributed to these lessons and who engaged in many extremely detailed lab meeting conversations about everything from normalization methods to plot colours.
Finally, be realistic about what questions your data can/can’t answer, and what your own biases are. Let the data speak for itself, after setting it up for success with a strong study design and robust analyses.
April 24, 2025 at 8:56 PM
Finally, be realistic about what questions your data can/can’t answer, and what your own biases are. Let the data speak for itself, after setting it up for success with a strong study design and robust analyses.
Next, stratification. Sometimes good/necessary (ex. when studying the X or Y chromosomes). However, when stratifying in other cases by sex, ancestry, etc., often better to replace or validate stratified analyses with interaction tests or other approaches.
April 24, 2025 at 8:56 PM
Next, stratification. Sometimes good/necessary (ex. when studying the X or Y chromosomes). However, when stratifying in other cases by sex, ancestry, etc., often better to replace or validate stratified analyses with interaction tests or other approaches.
Over the years we and others have identified several epigenetic signatures driven by underlying genetic variation. Consider the demographics of your study populations and of comparison groups, make sure things are balanced, and investigate these epi(genetic) hits when they arise!
April 24, 2025 at 8:56 PM
Over the years we and others have identified several epigenetic signatures driven by underlying genetic variation. Consider the demographics of your study populations and of comparison groups, make sure things are balanced, and investigate these epi(genetic) hits when they arise!
Also, variation associated with secondary variables like cell composition, sex, or genetic ancestry might itself be an interesting result. We’d encourage not always adjusting these effects away.
April 24, 2025 at 8:56 PM
Also, variation associated with secondary variables like cell composition, sex, or genetic ancestry might itself be an interesting result. We’d encourage not always adjusting these effects away.
An inconvenient truth of working with heterogeneous tissues is that sampling variation can influence your study. We demonstrate this by identifying cell composition variation across several public placental DNAme datasets. Something to consider when comparing studies!
April 24, 2025 at 8:56 PM
An inconvenient truth of working with heterogeneous tissues is that sampling variation can influence your study. We demonstrate this by identifying cell composition variation across several public placental DNAme datasets. Something to consider when comparing studies!
Next – with great power comes great responsibility. During study design and analysis, take care to identify and address sources of bias that can lead to false positives. And - because it’s impossible to avoid bias, replicate & validate findings in independent data.
April 24, 2025 at 8:56 PM
Next – with great power comes great responsibility. During study design and analysis, take care to identify and address sources of bias that can lead to false positives. And - because it’s impossible to avoid bias, replicate & validate findings in independent data.
Our recommendation is to check for aneuploidy and other genetic variation in your data & account for it in analyses where possible. Better yet – study these samples directly.
April 24, 2025 at 8:56 PM
Our recommendation is to check for aneuploidy and other genetic variation in your data & account for it in analyses where possible. Better yet – study these samples directly.
We begin by emphasizing that the placenta derives from the conceptus, but its development precedes and is largely independent of fetal development. This is important for ‘omics studies as genetic variation in the placenta (CPM, mutations) can influence both fetal and maternal health.
April 24, 2025 at 8:56 PM
We begin by emphasizing that the placenta derives from the conceptus, but its development precedes and is largely independent of fetal development. This is important for ‘omics studies as genetic variation in the placenta (CPM, mutations) can influence both fetal and maternal health.
The second major lesson I learned working in the Robinson lab was to consider the outliers. This lesson underlies much of the next few examples!
April 24, 2025 at 8:56 PM
The second major lesson I learned working in the Robinson lab was to consider the outliers. This lesson underlies much of the next few examples!
First, if working with @wprobins.bsky.social taught me anything, it’s that good research questions yield interesting results - whether positive (p < 0.05) or negative. With a solid research method, negative results add great value to the field.
April 24, 2025 at 8:56 PM
First, if working with @wprobins.bsky.social taught me anything, it’s that good research questions yield interesting results - whether positive (p < 0.05) or negative. With a solid research method, negative results add great value to the field.
This project was an odyssey and would not have been possible without support from @wprobins.bsky.social, Carolyn Brown, Allison Matthews (Cotton), and a productive collaboration with Melissa Wilson, Tanya Phung, and Seema Plaisier! Many thanks also to Maria Peñaherrera and the Roblab for support!
March 4, 2025 at 6:44 PM
This project was an odyssey and would not have been possible without support from @wprobins.bsky.social, Carolyn Brown, Allison Matthews (Cotton), and a productive collaboration with Melissa Wilson, Tanya Phung, and Seema Plaisier! Many thanks also to Maria Peñaherrera and the Roblab for support!
We still have yet to understand what this means for the sex differences phenotypically observed in pregnancy. Hopefully, time and future research will tell!
March 4, 2025 at 6:36 PM
We still have yet to understand what this means for the sex differences phenotypically observed in pregnancy. Hopefully, time and future research will tell!
My personal theory is because the placenta is (1) low-methylated across the entire genome and (2) arises largely from TE cells differentiating during the waves of epigenetic reprogramming, maybe it has come up with a unique strategy of achieving XCI without relying on DNAme.
March 4, 2025 at 6:36 PM
My personal theory is because the placenta is (1) low-methylated across the entire genome and (2) arises largely from TE cells differentiating during the waves of epigenetic reprogramming, maybe it has come up with a unique strategy of achieving XCI without relying on DNAme.
I'm fascinated by this result - what is keeping these low-methylated genes silent in placenta? Are other molecular players (histone marks, XIST, etc) more critically important than DNAme? Why does XCI appear to work differently here than in most other contexts?
March 4, 2025 at 6:36 PM
I'm fascinated by this result - what is keeping these low-methylated genes silent in placenta? Are other molecular players (histone marks, XIST, etc) more critically important than DNAme? Why does XCI appear to work differently here than in most other contexts?
Where low promoter DNAme usually indicates genes that escape XCI in other tissues, in the placenta most genes have low promoter DNAme, and yet they are not escaping according to gene expression evidence!
March 4, 2025 at 6:36 PM
Where low promoter DNAme usually indicates genes that escape XCI in other tissues, in the placenta most genes have low promoter DNAme, and yet they are not escaping according to gene expression evidence!
To test this theory we took gene-level XCI expression calls from placenta (Phung et al. 2022) and tested whether DNAme in our placentas corresponded with the XCI status of associated genes. It did (R~0.3)...but the relationship was weaker than is seen in other tissues!
March 4, 2025 at 6:36 PM
To test this theory we took gene-level XCI expression calls from placenta (Phung et al. 2022) and tested whether DNAme in our placentas corresponded with the XCI status of associated genes. It did (R~0.3)...but the relationship was weaker than is seen in other tissues!
And then the reason we started this project - as previous work has shown that promoter DNAme correlates with silencing on the inactive X, we wondered whether low DNAme = more escape from XCI in placenta??
March 4, 2025 at 6:36 PM
And then the reason we started this project - as previous work has shown that promoter DNAme correlates with silencing on the inactive X, we wondered whether low DNAme = more escape from XCI in placenta??
We explore several factors associated with X-linked DNAme variation, and see suggestions that cell composition is a major player. More work coming from @wprobins.bsky.social lab on this in the future! 🤓
March 4, 2025 at 6:36 PM
We explore several factors associated with X-linked DNAme variation, and see suggestions that cell composition is a major player. More work coming from @wprobins.bsky.social lab on this in the future! 🤓
We then demonstrate that the low DNAme status of the X chromosome is most extreme in XX (female) placentas, which implicates the inactive X. This low methylation is not explained fully by PMDs or repetitive elements, and is consistent across gestational ages (i.e. it wasnt methylated to begin with!)
March 4, 2025 at 6:36 PM
We then demonstrate that the low DNAme status of the X chromosome is most extreme in XX (female) placentas, which implicates the inactive X. This low methylation is not explained fully by PMDs or repetitive elements, and is consistent across gestational ages (i.e. it wasnt methylated to begin with!)
First, we find that the placenta (purple) is an outlier compared to other tissues by global X-linked DNAme profiles - in both sexes. This is already known on the autosomes but is relatively novel information for the X.
March 4, 2025 at 6:36 PM
First, we find that the placenta (purple) is an outlier compared to other tissues by global X-linked DNAme profiles - in both sexes. This is already known on the autosomes but is relatively novel information for the X.