Jeff Mold
@jeffmold.bsky.social
American/Swedish Biomedical Scientist studying immunology and cancer. My favorite cell atlases say “here be dragons” on the UMAPs. @karolinska institute
https://scholar.google.com/citations?hl=en&user=_owb98cAAAAJ&view_op=list_works&sortby=pubdate
https://scholar.google.com/citations?hl=en&user=_owb98cAAAAJ&view_op=list_works&sortby=pubdate
Skiing in Kosovo not bad!
February 3, 2026 at 12:26 PM
Skiing in Kosovo not bad!
Similarly it is not standard (to my knowledge) to try and interpret whether having a higher fluorescence for IFNg means that the cell is better at making it compared to lower fluorescence. Usually gates are drawn and cells are determined to be capable of responding to a stimulation or not.
January 22, 2026 at 9:28 AM
Similarly it is not standard (to my knowledge) to try and interpret whether having a higher fluorescence for IFNg means that the cell is better at making it compared to lower fluorescence. Usually gates are drawn and cells are determined to be capable of responding to a stimulation or not.
This is a random image from google - but it illustrates the point. Nothing can be inferred from the level of Ki67 expression here. A low expression might indicate background staining but generally it is treated as either a dividing or non-dividing cell. Being high or medium high is irrelevant.
January 22, 2026 at 9:28 AM
This is a random image from google - but it illustrates the point. Nothing can be inferred from the level of Ki67 expression here. A low expression might indicate background staining but generally it is treated as either a dividing or non-dividing cell. Being high or medium high is irrelevant.
So to come up with this figure (below) they first took every gene set from MSIGDB and then they filtered for ones with 'stem' in the name and at least 4 genes (which seems to be no filter because they have the same 1197 before and after). In addition they harvest signatures from selected papers.
January 22, 2026 at 9:18 AM
So to come up with this figure (below) they first took every gene set from MSIGDB and then they filtered for ones with 'stem' in the name and at least 4 genes (which seems to be no filter because they have the same 1197 before and after). In addition they harvest signatures from selected papers.
I have a serious question for the bioinformatics community - it is now very trendy to assign genes to modules or groups and then annotate those groups using various strategies. When I read the methods for papers it is unclear to me how one comes up with the strategy they use - see below:
January 22, 2026 at 9:18 AM
I have a serious question for the bioinformatics community - it is now very trendy to assign genes to modules or groups and then annotate those groups using various strategies. When I read the methods for papers it is unclear to me how one comes up with the strategy they use - see below:
Trump keeps complaining to the Norwegian government about his lack of prizes…. “https://edition.cnn.com/2026/01/19/europe/trump-norway-nobel-prize-snub-intl”
January 20, 2026 at 4:45 AM
Trump keeps complaining to the Norwegian government about his lack of prizes…. “https://edition.cnn.com/2026/01/19/europe/trump-norway-nobel-prize-snub-intl”
and (2) I never feel like the actual data looks all that different regardless of the statistics. The graphs don't look as if they suggest a difference really exists. In this paper www.nature.com/articles/s41...
They used a BACH2 regulon score on human CAR T cells - I am perplexed at the p-value
They used a BACH2 regulon score on human CAR T cells - I am perplexed at the p-value
January 17, 2026 at 3:46 PM
and (2) I never feel like the actual data looks all that different regardless of the statistics. The graphs don't look as if they suggest a difference really exists. In this paper www.nature.com/articles/s41...
They used a BACH2 regulon score on human CAR T cells - I am perplexed at the p-value
They used a BACH2 regulon score on human CAR T cells - I am perplexed at the p-value
The authors write this about the human data - I think it is quite a stretch to say the data above demonstrates a role for PARP in memory T cell differentiation in humans?!? Right? Or am I a jerk :) (or both)
January 13, 2026 at 4:01 PM
The authors write this about the human data - I think it is quite a stretch to say the data above demonstrates a role for PARP in memory T cell differentiation in humans?!? Right? Or am I a jerk :) (or both)
So in Ext Fig 3 they define the groups - and throughout the rest of the paper this comparison is made (Tcm vs CD44hiCD62Lhi) - wouldn't it at least be worth asking if the top percentile of untreated cells was =?
January 13, 2026 at 3:40 PM
So in Ext Fig 3 they define the groups - and throughout the rest of the paper this comparison is made (Tcm vs CD44hiCD62Lhi) - wouldn't it at least be worth asking if the top percentile of untreated cells was =?
A lot of the results are based on gating like this - where they take cells that are all in the top quadrant (presumably all CD44+CD62L+ and then divide them into CD44hiCD62Lhi vs intermediate and calculate lots of stats based on secondary FACS data often not shown (like IFNg, Ki67 etc..)
January 13, 2026 at 3:40 PM
A lot of the results are based on gating like this - where they take cells that are all in the top quadrant (presumably all CD44+CD62L+ and then divide them into CD44hiCD62Lhi vs intermediate and calculate lots of stats based on secondary FACS data often not shown (like IFNg, Ki67 etc..)
There is a bit of human data here and I would first like to comment on that - these are from ovarian cancer patients samples pre and post PARP treatment. It is a strange result for a few reasons I will go through: (1) why use CD62L MFI on cells already gated as CD45RO+CCR7+ and what is that axis?
January 13, 2026 at 3:40 PM
There is a bit of human data here and I would first like to comment on that - these are from ovarian cancer patients samples pre and post PARP treatment. It is a strange result for a few reasons I will go through: (1) why use CD62L MFI on cells already gated as CD45RO+CCR7+ and what is that axis?
In fact I would interpret the three subclusters of cells they see as basically roughly equivalent barring some extra protein based measurements which were more clear than the scRNA data. It clear in the bottom figure where we are looking at mostly high expressed genes with relative differences
January 13, 2026 at 2:15 PM
In fact I would interpret the three subclusters of cells they see as basically roughly equivalent barring some extra protein based measurements which were more clear than the scRNA data. It clear in the bottom figure where we are looking at mostly high expressed genes with relative differences
The idea of a novel 'GZMA+KLF2+' vs 'GZMA+KLRG1+' population of T cells is not really borne out of the data - Based on the dotplot the vast majority of cells are KLF2+ and there MIGHT be differences in expression levels between some populations but hard to say if it is meaningful.
January 13, 2026 at 2:15 PM
The idea of a novel 'GZMA+KLF2+' vs 'GZMA+KLRG1+' population of T cells is not really borne out of the data - Based on the dotplot the vast majority of cells are KLF2+ and there MIGHT be differences in expression levels between some populations but hard to say if it is meaningful.
100% Border Collies - ok now it makes sense.
January 9, 2026 at 1:57 PM
100% Border Collies - ok now it makes sense.
I am trying to figure out the breed of the gifted dogs - it looks like they are Border Collies from the cartoons, at least the controls are Border Collies. This passage from Ref24 of the paper stands out - If you enter your dog in a Scientific study please name it something like 'Pickles'...
January 9, 2026 at 1:54 PM
I am trying to figure out the breed of the gifted dogs - it looks like they are Border Collies from the cartoons, at least the controls are Border Collies. This passage from Ref24 of the paper stands out - If you enter your dog in a Scientific study please name it something like 'Pickles'...
I wasn't aware they were looking? Sounds like a George Church venture.
January 9, 2026 at 1:38 PM
I wasn't aware they were looking? Sounds like a George Church venture.
I know this one! It’s great. I’d say that there’s a decent amount of IgG though and I wonder how many PCs you need to inflict major damage in a tissue given how few are needed in BM to afford humoral immunity?
December 16, 2025 at 8:44 PM
I know this one! It’s great. I’d say that there’s a decent amount of IgG though and I wonder how many PCs you need to inflict major damage in a tissue given how few are needed in BM to afford humoral immunity?
Oh no... we have a new 'single cell RNAseq reveals...'
December 11, 2025 at 12:03 PM
Oh no... we have a new 'single cell RNAseq reveals...'
Also it’s a bit of a red flag when they get a result that goes against their hypothesis and they try to hand wave it away…
December 9, 2025 at 9:32 AM
Also it’s a bit of a red flag when they get a result that goes against their hypothesis and they try to hand wave it away…
Just 8 months ago he was heart breakingly cute
December 3, 2025 at 1:14 PM
Just 8 months ago he was heart breakingly cute
I just wanna brag about what a beautiful little puppy I have :)
December 3, 2025 at 1:05 PM
I just wanna brag about what a beautiful little puppy I have :)