David Rach
@davidrach.bsky.social
Immunology PhD student @UMBaltimore. Views are my own. Favorite rabbit holes include #flowcytometry #rstats #bioconductor, with dashes of #coffee #kayaking #birding #ultrunning #languages. RPCV Ghana 16->18, TCK Mexico 96->11 https://github.com/DavidRach/
Happy to officially announce that we are hosting a free “Cytometry in R” weekly mini-course, aimed at coding beginners, both in-person here at the University of Maryland, Baltimore and virtually, starting up in December/January. For details: umgccfcss.github.io/CytometryInR/ #rstats #flowcytometry
November 5, 2025 at 9:36 PM
Happy to officially announce that we are hosting a free “Cytometry in R” weekly mini-course, aimed at coding beginners, both in-person here at the University of Maryland, Baltimore and virtually, starting up in December/January. For details: umgccfcss.github.io/CytometryInR/ #rstats #flowcytometry
And as testified by the sheer number of Main and Supplementary Figures, I really enjoyed retrieving the manually-appended gating information to generate these #beeswarm #boxplots #ggplot2 showing each individuals proportion of positive cells for every marker by exposure group. 16/n
August 25, 2025 at 4:44 AM
And overall we found few significant differences between cord blood ILTs of HIV unexposed and HIV-exposed neonates. Those we did find were primarily in the CD4- CD8- Vd2 T cells, with HEU neonates that had greater viral exposure before birth showing higher frequencies of CD62L and IFNg. 15/n
August 25, 2025 at 4:44 AM
And overall we found few significant differences between cord blood ILTs of HIV unexposed and HIV-exposed neonates. Those we did find were primarily in the CD4- CD8- Vd2 T cells, with HEU neonates that had greater viral exposure before birth showing higher frequencies of CD62L and IFNg. 15/n
We saw some interesting overlaps in phenotype and functional marker expression between cord blood NKT and CD4+ Vd2 T cells, as well as MAIT and CCR6+ Vd2 T cells. 14/n
August 25, 2025 at 4:44 AM
We saw some interesting overlaps in phenotype and functional marker expression between cord blood NKT and CD4+ Vd2 T cells, as well as MAIT and CCR6+ Vd2 T cells. 14/n
Leaving the "What defines a cluster? How many clusters are there? And did I generate them into a shape of a coffee cup with Picasso?" discussion for another day, within our 29-color panel, the dimensionality visualization tracked with overall marker co-expression data. 13/n
August 25, 2025 at 4:44 AM
Leaving the "What defines a cluster? How many clusters are there? And did I generate them into a shape of a coffee cup with Picasso?" discussion for another day, within our 29-color panel, the dimensionality visualization tracked with overall marker co-expression data. 13/n
Cord blood ILT were indeed rare, with Vd2+ T cells making up 0.7% of T cells. 11/n
August 25, 2025 at 4:44 AM
Cord blood ILT were indeed rare, with Vd2+ T cells making up 0.7% of T cells. 11/n
And finally, not wanting to "just blindly trust the tSNE plot", I extended #bioconductor #cytoverse R package functions to append individual cells with manual-gating information before they were concatenated together. This allowed back-calculating proportion positive cells for each individual. 9/n
August 25, 2025 at 4:44 AM
And finally, not wanting to "just blindly trust the tSNE plot", I extended #bioconductor #cytoverse R package functions to append individual cells with manual-gating information before they were concatenated together. This allowed back-calculating proportion positive cells for each individual. 9/n
Having had a taste of "coding my way out of a untractable problems", I was hooked. For this paper, I used #rstats for the analysis and figures, wrote the manuscript with #quarto and #git, assembled final figures with #inkscape, uploaded .fcs file data to #ImmPort and made my code #opensource. 8/n
August 25, 2025 at 4:44 AM
Which is where my contribution to the project began, and marks the start of the #flowcytometry and #rstats adventures! Sporadically for certain experiments or specimens, we encountered unmixing errors! Back in 2022, identifying problems involved time-intensive searching through OneByN plots 6/n
August 25, 2025 at 4:44 AM
Which is where my contribution to the project began, and marks the start of the #flowcytometry and #rstats adventures! Sporadically for certain experiments or specimens, we encountered unmixing errors! Back in 2022, identifying problems involved time-intensive searching through OneByN plots 6/n
Working with our Malawian colleagues, we recruited a maternal-infant cohort to better understand how maternal viral load and antiretroviral initiation before birth affects neonatal immune response. Cord blood mononuclear cells were cryopreserved in liquid nitrogen, and shipped to to Baltimore. 5/n
August 25, 2025 at 4:44 AM
Working with our Malawian colleagues, we recruited a maternal-infant cohort to better understand how maternal viral load and antiretroviral initiation before birth affects neonatal immune response. Cord blood mononuclear cells were cryopreserved in liquid nitrogen, and shipped to to Baltimore. 5/n
Due to increased access to #antiretrovirals before birth, HIV-exposed but uninfected (HEU) infants are a growing clinical population. However, they experience more lower respiratory and diarrheal infections during first 6 months of life than their HIV-unexposed counterparts. Why remains unclear 3/n
August 25, 2025 at 4:44 AM
Due to increased access to #antiretrovirals before birth, HIV-exposed but uninfected (HEU) infants are a growing clinical population. However, they experience more lower respiratory and diarrheal infections during first 6 months of life than their HIV-unexposed counterparts. Why remains unclear 3/n
"Note to self: Definitely need to rethink how I wrote that particular function" 😅 (Second kernel initiated shutdown of #positron after all available RAM was used up mid map).
August 19, 2025 at 12:58 AM
"Note to self: Definitely need to rethink how I wrote that particular function" 😅 (Second kernel initiated shutdown of #positron after all available RAM was used up mid map).
And first of the fun times, when attempting to login using #Wayland I get to a frozen #KDE screen. Interestingly enough, when selecting #X11 this doesn't happen. Anyone have any thoughts on the cause/problem-solving? Fortunately have time to tinker this weekend 🙂 #Debian #Trixie #Linux
August 9, 2025 at 9:43 PM
August 9, 2025 at 8:48 PM
Solved!!! Credit and thanks to @stephan-koenig.bsky.social for spotting it and replying on Data Science Learning Community Slack (@dslc.io). My quarto render action had a "with: to: html" in it that was responsible for the altered website render, when removed, back to normal 😅❗❤️ #rstats #Quarto 4/n
May 2, 2025 at 2:46 AM
Solved!!! Credit and thanks to @stephan-koenig.bsky.social for spotting it and replying on Data Science Learning Community Slack (@dslc.io). My quarto render action had a "with: to: html" in it that was responsible for the altered website render, when removed, back to normal 😅❗❤️ #rstats #Quarto 4/n
Current GitHub actions .yaml can be found here: github.com/DavidRach/Ba... Thanks in advance for any insight that you can point me towards! David 3/3
May 1, 2025 at 12:52 AM
Current GitHub actions .yaml can be found here: github.com/DavidRach/Ba... Thanks in advance for any insight that you can point me towards! David 3/3
When rendered via Github actions, the quarto-dashboard folder gets deleted and replaced by a tabset folder (commit changes attached). Obviously missing something required from the GitHub actions .yaml to succeed, but not quite clear on the what 2/3
May 1, 2025 at 12:52 AM
When rendered via Github actions, the quarto-dashboard folder gets deleted and replaced by a tabset folder (commit changes attached). Obviously missing something required from the GitHub actions .yaml to succeed, but not quite clear on the what 2/3
Hey #Quarto #RStats friends, my Quarto website when rendered on my local computer correctly places the plotly widgets into rows and columns (creating a site_lib/quarto-dashboard folder with .js files in the process). However, when I set up a #githubaction it gets rendered single column🙃 Any ideas?
May 1, 2025 at 12:52 AM
Hey #Quarto #RStats friends, my Quarto website when rendered on my local computer correctly places the plotly widgets into rows and columns (creating a site_lib/quarto-dashboard folder with .js files in the process). However, when I set up a #githubaction it gets rendered single column🙃 Any ideas?
Can't even remember what the original coding project goal I set out with for this weekend was, but every single thing code related descended into yak shaving not one but an entire herd 😮💨
images.app.goo.gl/7q242s9B46q3...
images.app.goo.gl/7q242s9B46q3...
April 14, 2025 at 2:32 AM
Can't even remember what the original coding project goal I set out with for this weekend was, but every single thing code related descended into yak shaving not one but an entire herd 😮💨
images.app.goo.gl/7q242s9B46q3...
images.app.goo.gl/7q242s9B46q3...
Extra #autofluorescence (s), one hell of a drug😅 34-color panel, gated on live lymphocytes. Unmixed on the Aurora, left row with general AF extract, right column with an additional AF tag. Main AF had primary peak V7 detector, second AF (25% live cells) main peak was UV7 #spectral #flow #cytometry
March 18, 2025 at 1:43 AM
Extra #autofluorescence (s), one hell of a drug😅 34-color panel, gated on live lymphocytes. Unmixed on the Aurora, left row with general AF extract, right column with an additional AF tag. Main AF had primary peak V7 detector, second AF (25% live cells) main peak was UV7 #spectral #flow #cytometry
Guess who while attempting to fix the #Wine v10 update issue that broke #FlowJo on his #Debian #Linux boot ended up screwing up #Wayland and #KDE enough that SDDM wouldn't recognize hitting enter on the password screen? 😅 Managed to reinstall from terminal and now back to normal, but uffda! 🐧😂
February 9, 2025 at 5:43 AM
My first foray into #rustlang, using extendr to speed up a few things in one of my #cytometry #rstats packages. End result, l learned a bunch and a clear improvement over the R equivalent that's running Fortran. TBD if it throws an additional note for #Bioconductor when running BiocCheck later 😂
January 18, 2025 at 9:45 PM
My first foray into #rustlang, using extendr to speed up a few things in one of my #cytometry #rstats packages. End result, l learned a bunch and a clear improvement over the R equivalent that's running Fortran. TBD if it throws an additional note for #Bioconductor when running BiocCheck later 😂
And Fluorophore BUV563 (CD69 as antigen in this case) somewhere in between.
And that's where I leave this fun rabbit-hole for now and return to what I actually need to get done this AM. 5/🧵
And that's where I leave this fun rabbit-hole for now and return to what I actually need to get done this AM. 5/🧵
December 17, 2024 at 10:58 AM
And Fluorophore BUV563 (CD69 as antigen in this case) somewhere in between.
And that's where I leave this fun rabbit-hole for now and return to what I actually need to get done this AM. 5/🧵
And that's where I leave this fun rabbit-hole for now and return to what I actually need to get done this AM. 5/🧵