Rethinking disomy: Autosomal expression bias

Different forms of monoallelic gene expression are observed in disomic organisms. Imprinting is perhaps the best known (Reik and Walter, 2001). Imprinting is heritable and considered to be a form of monoallelic expression given that the expression of one allele is completely shut off in every cell in the body (Fig. 1 A). X-chromosome inactivation, in which one of the two X chromosomes is mostly inactivated via Xist, affecting most if not all cells in females (Disteche and Berletch, 2015; Schulz and Heard, 2013) is also a form of monoallelic expression (Fig. 1 B). A given cell can also present with as discontinuous transcription resulting in transcriptional bias toward one allele and then switch the bias to the other allele, generating a form of monoallelic expression that is relatively transient. This is known as transcriptional bursting (Fig. 1 C) (Tunnacliffe and Chubb, 2020). This review will not cover imprinting (Reik and Walter, 2001) and X-chromosome inactivation (Disteche and Berletch, 2015; Schulz and Heard, 2013), but it will compare and contrast with transcriptional bursting concepts (Tunnacliffe and Chubb, 2020). This review will not cover B cell receptors, T cell receptors, olfactory receptors, or cadherins, all of which undergo defined monoallelic forms of autosomal gene expression. Instead, I will focus here on a type of monoallelic expression more commonly referred to as autosomal random monoallelic expression (aRMAE) (Chess, 2016; Gendrel et al., 2016; Kanata et al., 2024; Khamlichi and Feil, 2018; Stewart et al., 2025), in which the asymmetric expression of one allele is documented, but occurs somatically and is mitotically stable, resulting in de facto transcriptional mosaicism (Fig. 1 D). In fact, the situation is even more complicated, as aRMAE is technically a misnomer. Expression is rarely exclusively from one allele, but on a continuum (Fig. 2 A), and while it is mitotically stable, this commitment need not be permanent exclusively (Fig. 2 B). Consensus allelic imbalance of >80% is required for expression to be considered monoallelic currently (Xu et al., 2017). This concept and these definitions merit further discussions and would benefit from a unifying all-encompassing framework.

The following terminology including monoallelic expression, asymmetric expression, autosomal expression bias, autosomal asymmetric expression, aRMAE, widespread dynamic RME, and widespread fixed RME are all terms widely used to describe what we will here call “autosomal random monoallelic expression (aRMAE).”

The first case of aRMAE to be identified was probably that of the albumin locus (Michaelson, 1993). Michaelson obtained sera directed against one of the two allelic forms of albumin and demonstrated that most of the cells expressing albumin in the liver of heterozygous Alb-1a x Alb-1c mice expressed only one of the two alleles (Michaelson, 1993). In 1998, a flurry of papers documenting aRMAE for the IL-2 (Hollander et al., 1998) and IL-4 (Bix and Locksley, 1998; Rivière et al., 1998) loci were published, followed by articles documenting aRMAE for the IL-3, IL-5, IL-13 (Kelly and Locksley, 2000), and IL-10 (Calado et al., 2006) loci.

One good example of aRMAE is provided by a manuscript published in 1998 by Isabelle Rivière , Mary Jean Sunshine, and Dan Littman. These researchers applied the cutting-edge technology of the time to improve our understanding of T cell development and of the newly discovered Th1 vs. Th2 polarization (Rivière et al., 1998). They did this by producing a mouse in which one of the two copies of the IL-4 gene expressed a cDNA encoding the human CD2 (huCD2) cell-surface molecule, whereas the other expressed IL-4. The idea was, of course, to use this surface huCD2 expression to facilitate the tracking of Th2 cells. However, they found that contrary to expectations, some T cells expressed huCD2 but did not produce IL-4, others produced IL-4 but did not express huCD2, and others, were, as expected, double-positive for huCD2 and IL-4. The manuscript relating this discovery, and those preceding and following it, suggested that stochastic events caused this phenomenon at these interleukin loci, leading to development of the concept of “transcriptional bursting” (when one allele is dominantly expressed at a given time point, but the cell concerned can switch to biallelic expression).

However, even the evidence presented in this article published in 1998 indicated that aRMAE was probably a nuanced, complex, and highly regulated process worthy of further exploration. The authors showed that (1) the strength of the T cell receptor signal was a determinant of monoallelic expression, (2) some T cell clones had a stable aRMAE commitment despite repeated stimulation in long-term culture, and (3) some clones reverted to biallelic expression or underwent a complete switch to the other allele during the course of the experiment, a phenomenon now classified as transcriptional bursting. These experiments were sufficient in themselves to suggest a mechanistic complexity of aRMAE.

The following year (1999), groups from Sweden and the UK showed that known imprinted genes could also display monoallelic expression in the absence of discriminating parental marks (Ohlsson et al., 1999). They examined human androgenetic cells, which cannot discriminate between alleles as both alleles are of paternal origin. An analysis of dispermic complete hydatidiform moles (empty eggs, each fertilized by two sperm) showed that IGF2 and H19 (previously shown to be imprinted) could be transcriptionally active in a variegated manner, resulting in aRMAE. In 2003, Allen et al. (2003) suggested that genes displaying aRMAE were preceded by higher densities of LINE-1 sequences, less-truncated LINE-1 elements, fewer CpG islands, and fewer base pairs of short interspersed nuclear element sequences than genes displaying biallelic expression. These patterns did not appear to cause aRMAE on their own but may have contributed to the priming of the locus for asymmetric transcription. These findings suggest that a gene positioning pattern may act as a determinant of propensity for monoallelic expression, but this remains a matter of debate.

18 years ago, Gimelbrant et al. discovered that aRMAE is relatively widespread and not restricted to select loci (Gimelbrant et al., 2007). This was experimentally identified in human B-lymphoblastoid cell lines through the derivation of clonal cell lines by single-cell cloning. Gimelbrant et al. suggested that almost 10% of assessable genes (expressed and with at least one SNP, making it possible to differentiate between the maternal and paternal alleles) could undergo aRMAE. 5 years later, this was also documented in clonal mouse cell lines from heterozygous mouse strains. The authors concluded that aRMAE was widespread throughout the genome (over 10% of genes underwent aRMAE), was not subject to chromosome-wide coordination, and varied between cell types (Zwemer et al., 2012).

In 2014, the first study to apply newly developed single-cell RNA-sequencing technology (smart-seq) to murine cells also documented widespread (12–24%) aRMAE. Like most other studies, this study was unable to identify any clear expression patterns and reported appreciable levels of variation in closely related embryonic cells and mature murine liver cells (Deng et al., 2014). Like a subsequent study published in 2016 (Reinius et al., 2016), it suggested that most of the discoveries reported could probably be attributed to transcriptional bursting rather than mitotically stable aRMAE. As reported for the initial discovery of aRMAE at the IL-4 locus, it is possible that the same locus can undergo transcriptional bursts, switching from one allele to the other, temporarily committing to the biased expression of one allele, or displaying stable aRMAE. These notions concerning the plasticity of aRMAE have yet to be fully demonstrated, but there is evidence to suggest that they are plausible (Rivière et al., 1998).

A study performed in 2019 compared rates of aRMAE between heterokaryotypic monozygotic (MZ) twins, homokaryotypic MZ twins, and unrelated pairs of nontwins. The degree of aRMAE discordance in MZ twins (2.7%) did not differ between heterokaryotypic and homokaryotypic MZ twins but was about a 10th that expected between pairs of unrelated, male, or female nontwins. For me, this suggests that at least some of the propensity for aRMAE is germline-encoded and unlikely to be stochastic (da Silva Francisco Junior et al., 2019). In 2025, Luong et al. suggested that genetic variation does indeed skew the aRMAE of Pvt1 via differential TFAP2a binding (Luong et al., 2025). There is therefore evidence to suggest that there is at least some heritable and probably acquired genetic control of aRMAE.

A study performed in 2023 investigated aRMAE in all human tissues (GTEx datasets for 832 people and 54 tissues). It used a very elegant statistical model of transcriptional dispersion as a proxy for the presence or absence of aRMAE in tissue samples subjected to bulk RNA sequencing. This sophisticated proxy for aRMAE made it possible to show that all human tissues were prone to aRMAE. Based on the study of patterns of transcriptional dispersion, the authors suggested that these processes were probably controlled by several different mechanisms (Kravitz et al., 2023).

In 2025, we used a clonal system based on primary T cells from nine healthy New Yorkers to show that mitotically stable aRMAE could, indeed, be observed in the immune system of adult humans and affected about 5% of the human genome, again with no identification of any clear genomic positions or gene classes associated with this process (Stewart et al., 2025). We also showed that aRMAE could, indeed, influence disease outcome in monogenic disorders with incomplete penetrance. Some patients displaying genetic heterozygosity in all cells displayed transcriptional homozygosity in certain tissues due to aRMAE, which prompted us, at least for genetic diagnosis, to distinguish between “genotype” and “transcriptotype.”

All these studies, and many more discussed below, suggest that the biallelic expression of autosomal genes is not the only canon. Many genes undergo a transient monoallelic transcriptional burst, subsequently switching to a transcriptional burst off the other allele. Many genes are prone to mitotically stable aRMAE, which can occur in all tissues tested. It could even be argued that the cutoff for aRMAE could be set at 75% or 65% as opposed to 80%, or that it could be a continuum. If this were the case, it could even be suggested that most of the genome undergoes aRMAE, and that this is how biology works. This is a plausible hypothesis, but it remains to be proved.

aRMAE has perhaps been best studied in development. Stem cells have been shown to display aRMAE. In 2009, a group led by Andrew Chess, who also led the team that first showed aRMAE to be widespread (Gimelbrant et al., 2007), showed that allele-specific replication occurs in human embryonic stem cells (ESCs), preferentially in aRMAE genes (Dutta et al., 2009). Another group investigated the dynamics of monoallelic expression during development in clonal populations of hybrid mouse ESCs and neural progenitor cells (NPCs). They identified 67 aRMAE genes in ESCs and 376 aRMAE genes in NPCs, suggesting that there is a 5.6-fold increase in aRMAE rates upon differentiation (Eckersley-Maslin et al., 2014). Edith Heard and her group identified several hundred aRMAE genes in clonal NPC lines derived from ESCs. They noted that aRMAE occurred during differentiation and that the monoallelic state could be very stable once established. They suggested that if a mutation is present, it could lead to disease (Gendrel et al., 2014).

Others suggested that aRMAE in lymphocytes is mostly established after the HSC stages of development (Kubasova et al., 2022) and that loci previously characterized by aRMAE generally undergo a reset to the biallelic state in induced pluripotent stem cells (iPSCs) (Jeffries et al., 2016). One study identified risk loci via allele-specific open chromatin assessment in iPSC-derived neurons and suggested that aRMAE may, indeed, influence disease (Zhang et al., 2020). Just this year, Emily Mace’s group demonstrated for the first time that the aRMAE of a mutated allele can occur during the differentiation of iPSCs to NK cells. Two siblings with compound-heterozygous mutations in the GINS4 gene displayed differences in disease expressivity (a form of partial incomplete penetrance in which one member of a family is sicker than another despite both carrying the same mutations). Dr. Mace’s group showed that the GINS4 gene in the most affected sibling underwent aRMAE during differentiation into NK cells and expressed only the more severely affected null allele, whereas the GINS4 gene of the least affected sibling retained biallelic expression throughout the differentiation of iPSCs into NK cells (Seo et al., 2025). These results not only confirmed that aRMAE can occur during development, is stable, and is linked to disease penetrance (Stewart et al., 2025), but also suggested that the propensity to aRMAE may, indeed, be genetically encoded, as the iPSCs from one sibling repeatedly displayed aRMAE during NK cell differentiation, whereas those from the other sibling did not.

aRMAE has probably been insufficiently studied in cancer. In 2008, a study on lymphoblastoid cell lines confirmed the widespread presence of aRMAE (Tan et al., 2008). It provided evidence that almost 20% of the genome can undergo aRMAE. Perhaps more importantly, it evaluated aRMAE in lymphoblastoid cell lines derived from controls and patients with familial pancreatic cancer. Overall, aRMAE rates were similar between groups, but stronger aRMAE—referred to by the authors as “extreme aRMAE” (to indicate an even stronger shift toward [nearly 100%] the expression of a single allele)—in genes known to be associated with pancreatic cancer, such as BRCA2, FANCA, FANCD2, and PTCH1, and new candidate genes, such as BARD1, CDH1, and NBN, was found to occur in cell lines from patients with familial pancreatic cancer. Notably, in one patient, the authors found extreme aRMAE for a BRCA2 SNP (rs144848) known to increase pancreatic cancer risk significantly. In 2017, a meta-analysis of 40 studies from 30 publications showed that the rs144848 polymorphism was associated with non-Hodgkin lymphoma and that, if stratified by study design, the allele model was associated with breast cancer risk (Li et al., 2017). However, aRMAE was not assessed. It is possible that the associations detected would have been even stronger had expression data been available. Furthermore, one study showed that MAE rates were higher in tumor samples than in normal tissues and that they increased with tumor stage (Walker et al., 2012).

The notion that germline and acquired mutations that confer a predisposition to, cause, or promote the progression of cancer can also undergo aRMAE would open up new possibilities for assessing cancer risk and new treatment options. For example, if the DNA of a patient’s cancer contains V600E BRAF mutations, the patients are currently prescribed a drug such as vemurafenib to target this mutation. Assessment of the aRMAE of this variant in a cancer might reveal that the mutated allele is silenced, then the benefit of continued use of this drug would need reassessment, in this hypothetical situation. We might also find that expression is biallelic. In such cases, treatment would begin but we would monitor the stability of expression, thereby assessing the efficacy of the drug. This approach should provide an early enough warning that the response to the drug is waning (development of resistance), making it possible to act more rapidly. There are many examples of ways in which aRMAE can be used to guide the diagnosis and treatment of various forms of cancer.

DNA methylation is probably the most influential regulator of aRMAE, especially if strictly defined as stable and mitotically heritable (da Rocha and Gendrel, 2019). A correlation has been found between aRMAE and DNA methylation levels (Jeffries et al., 2012; Marion-Poll et al., 2021; Stewart et al., 2025), but this correlation does not hold for many documented aRMAE genes, even in clones of the same gene from the same individual (Stewart et al., 2025). A few studies have reported numerous examples of both an absence of correlation and strong correlation between monoallelic expression and DNA methylation (Aseem et al., 2013; Eckersley-Maslin et al., 2014; Gendrel et al., 2014; Gutierrez-Arcelus et al., 2013). However, prevailing evidence suggests that DNA methylation plays a role in aRMAE (Fig. 3 A) but is unlikely to be the only mechanism governing this type of plastic transcriptional regulation.

Histone modification is another well-accepted epigenetic mechanism leading to aRMAE (Fig. 3 A). In 2013, Nag et al. reported in a seminal paper that the presence of a chromatin signature for both active transcription (H3K36me3) and silencing (H3K27me3) within the gene body suggests that one allele is more active than the other (Nag et al., 2013). We have also implicated H3K27me3 in a mechanism governing aRMAE (Stewart et al., 2025). Recently, studies in murine models have confirmed the role of active histone marks, such as H3K27ac, H3K4me2, and H3K4me3, and repression marks, such as H3K9me3, H2AK199ub, and H3K27me3, along the polycomb complex in aRMAE (Luong et al., 2025). One study detected no association of aRMAE with individual alleles with both active (H3K4me3) and silent (H3K27me3) chromatin modifications (Thomas et al., 2011). Thus, while imperfect, there is clear experimental and logical evidence to suggest that histone modifications control allele-specific expression levels for at least some loci.

The role of enhancers in aRMAE is only just beginning to be investigated. In 2022, a study suggested that aRMAE results from gene regulation by enhancers rather than an aRMAE-specific epigenetic program, based on the study of aRMAE for NKG2D in murine NK cells (Kissiov et al., 2022). This concept is very interesting as it suggests that the cis or even perhaps trans distal regulation of transcription is the key to at least initiating biased transcription from one allele (Fig. 3 B). This notion is borrowed from studies of eQTL reporting that distal cis-acting single-nucleotide polymorphisms (inborn or acquired) may influence the transcriptional balance of the two alleles.

Another less well-studied concept is that of RNA stability. One study investigated how natural antisense transcripts (NATs)—RNA molecules that bind to and regulate other RNAs—contribute to the control of gene expression in mice. The authors found that an overlap of at least 29 base pairs between sense and antisense transcripts was required for efficient processing into endogenous small interfering RNAs (endo-siRNAs) within the nucleus, and that the orientation of these endo-siRNAs was tissue-specific. In the presence of a heterozygous genetic variant, there was a significant correlation between NATs and aRMAE. It is thought that NATs and their derived endo-siRNAs can regulate gene expression posttranscriptionally by influencing allele-specific gene silencing in response to genetic variation (Carlile et al., 2009). This principle has been insufficiently studied, and additional experimental evidence would be welcome (Fig. 3 C).

One report suggested Smchd1, an epigenetic modifier essential for X-chromosome inactivation, may be involved in controlling the expression of several autosomal gene clusters subject to monoallelic expression (imprinted genes, protocadherins) rather than having a role restricted to X inactivation (Mould et al., 2013). Other forms of monoallelic expression may also therefore have complex and overlapping mechanisms.

In 2017, a study suggested that there was an enrichment in random monoallelically accessible (RAMA) elements—regulatory elements accessible on only one allele in a given cell—in the promoters of aRMAE genes that might govern at least some monoallelic expression. Allelic choice at RAMA elements was stable across cell generations. This study suggested that while allelic choice at most RAMA elements was consistent with a stochastic process, up to 30% of RAMA elements deviated from these expectations, suggesting a more regulated, but as yet still unknown, mechanism (Xu et al., 2017).

MeCP2, a regulator of neuronal function and development, was recently shown to influence about 50 genes prone to aRMAE, probably contributing to diverse sets of mechanisms affecting aRMAE (Brousseau et al., 2020). Similarly, another study suggested that aRMAE genes are sorted and insulated from biallelically expressed genes via CTCF-mediated chromatin loops (Fig. 3 D), a structure via which the CTCF protein is thought to orchestrate genome-wide insulations of chromatin states through higher order chromatin organization (Chandradoss et al., 2020).

None of the documented and postulated mechanisms is sufficient in itself to explain the widespread mechanism by which aRMAE is established. However, considering all the possible mechanisms together, a picture of a complex, nuanced, and highly regulated process is beginning to emerge.

The first indirect evidence of relevance to human disease, albeit in the context of imprinted genes and their biology, was provided by the deletion of the KANK1 gene (also called ANKRD15), which is associated with neurodevelopmental diseases such as congenital cerebral palsy, hypotonia, quadriplegia, and intellectual disability. It was suggested that the inheritance pattern in a four-generation family was due to maternal imprinting, as all affected individuals inherited the deletion from their fathers and monoallelic protein expression was observed. However, there was an exception to the pattern of a paternally inherited deletion in the proband’s unaffected sibling inconsistent with maternal imprinting (Vanzo et al., 2013). It remains unknown whether this was due to aRMAE, but this remains a plausible hypothesis.

A study in 2013 investigated the relationship between aRMAE genes previously identified in clonal neural stem cells and neurodevelopmental disorders, such as autism and schizophrenia. Schizophrenia risk candidate genes identified by genome-wide association studies on the genetic association database were overrepresented among aRMAE genes (Jeffries et al., 2013). No material was available in this study to document this proposed mosaic transcriptional bias, but the findings obtained pointed to a possible epigenetic influence on genetic susceptibility. In the absence of documentation in humans, Gui et al. (2017) suggested that, as reported for aRMAE for the Tbx5 gene in mice, variable penetrance in humans with TBX5 mutations may potentially be explained by aRMAE.

My group studies inborn errors of immunity (IEIs), a group of Mendelian disorders caused by genetic defects of the innate or adaptive immune system. IEIs can present as a high degree of susceptibility to pathogens, autoinflammation, allergy, autoimmunity, and/or cancers. They may be inherited in a dominant or recessive, autosomal, or X-linked manner. However, genetic segregation is often imperfect, and the prevalence of incomplete penetrance is high, currently estimated at 20–40% (Bogunovic, 2025). We studied a patient with a mosaic gain-of-function mutation in JAK1 (S703I). This individual was about 50% mosaic in all tissues tested. We therefore performed single-cell RNA sequencing with the aim of genotyping the patient from RNA and then comparing the transcriptomes of WT/WT and WT/S703I cells. It was at this point that we noticed the mismatch between genotype and what we termed transcriptotype, as some cells expressed only S703I, despite being genetically WT/S703I (Gruber et al., 2020). It was this serendipitous discovery that prompted us to hypothesize that aRMAE might be one of the mechanisms governing this observation and influencing disease penetrance in other families.

Half a decade later, we have, indeed, shown that aRMAE is widespread in healthy individuals and that aRMAE is the cause of incomplete penetrance in a number of families with monogenic IEIs (Stewart et al., 2025). We first documented aRMAE in a family with a different JAK1 gain-of-function mutation (S700N). We observed transcriptional suppression of the mutated allele across cell types in a healthy carrier from the family, whereas the proband, in whom the mutated allele was also suppressed in most cells, displayed biallelic expression in T cells. This suggests that expression in T cells alone is sufficient for disease (Stewart et al., 2025). Similarly, we documented complete suppression of the mutated allele in a family carrying a dominant-negative STAT1 mutation. The healthy father was heterozygous for the dominant allele at the DNA level, whereas transcriptotype revealed the presence of only the WT allele in all cell types tested. This finding contrasted with those for his sick child, who was also heterozygous for the dominant allele at the DNA level in all cells and displayed biallelic transcript expression in monocyte subsets (but with suppression of the mutated allele in T cells) (Stewart et al., 2025). We also studied two cousins carrying heterozygous PLCG2 mutations but whose B cells differed in terms of their ability to produce antibodies. Digital-droplet PCR was performed after sorting B cells from the two cousins based on their calcium flux. Flux-negative B cells expressed only the mutated allele from the cousin who did not make antibodies, whereas flux-negative B expressed both the mutated allele and the WT allele stemming from the cousin who did make antibodies (Stewart et al., 2025). Similarly, we observed a direct correlation with aRMAE in a family with a mutation in CARD11 displaying incomplete penetrance for an IEI (Stewart et al., 2025). Soon after, as discussed above, the group of Emily Mace identified aRMAE as the cause of incomplete penetrance in siblings with mutations in GINS4 (Seo et al., 2025). More recently, a group from Denmark showed that aRMAE was responsible for the incomplete penetrance of CNS herpesvirus infections in a family with IRF7 mutations (Werner et al., 2025). Together, these discoveries demonstrating that, as previously postulated, aRMAE is a mechanism of incomplete penetrance, suggest that this is probably only the tip of the iceberg and that perhaps many other genetic diseases in which incomplete penetrance is common may benefit from the assessment of aRMAE in discordant family members.

aRMAE is still considered an exception, rather than the rule. This view was understandable until recently based on current norms concerning what constitutes aRMAE and the evidence acquired. Perhaps the first aspect that we need to address is that very few loci have been shown to display stringent monoallelic expression. Most genes classified as aRMAE genes have been shown to display true transcriptional bias, with either paternal or maternal expression, with over 80% of expression from one allele (Deng et al., 2014; Gimelbrant et al., 2007; Reinius et al., 2016; Stewart et al., 2025). This asymmetric pattern of transcription corresponds to a continuum extending from 80% upward, but rarely reaching 100%. The idea that aRMAE should be defined as >80% of transcripts coming from a single allele is supported by empirical and statistical rigor (Xu et al., 2017), but there is room for discussion. This standard may be too rigorous.

First, the technology for identifying aRMAE has improved, making it easier to track transcriptional activity more precisely at a much finer scale (Fig. 2 A). Second, current estimates that 5–15% of the human genome can undergo aRMAE would probably skyrocket to over 50% of the human genome if the threshold were moved to a transcriptional bias of 60%. Good but imperfect evidence in favoring of lowering the threshold is supplied by our measurement of allelic bias and its association with disease in the family with the JAK1 S700N mutation, in which this bias was well below 80% threshold. This example alone suggests that the current threshold should be revised. Furthermore, as shown even in the 1998 paper on the IL-4 locus, the line between transcriptional burst and aRMAE may be somewhat blurred. Strictly speaking, we currently think of transcriptional burst as the ability of a cell to transcribe a message from one and then from the other allele at different time points without undergoing mitosis. Even if the 80% threshold is retained as the rule, what happens if there is a prolonged period (e.g., 2 wk) of transcriptional bias toward one allele, but then a reversion to biallelic expression (Fig. 2 B). Should this be considered transcriptional burst or short-term aRMAE? On top of this, mitotically stable but reversible and irreversible aRMAE may occur. Perhaps considering the framework of aRMAE as a continuum would make it possible to suggest and even prove that aRMAE is highly plastic in human biology and beyond, and that this is the norm rather than an exception.

We and others have postulated (Gendrel et al., 2016; Gruber and Bogunovic, 2020; Gruber et al., 2020; Gui et al., 2017; Kravitz et al., 2023; Marion-Poll et al., 2021) and recently demonstrated (Stewart et al., 2025) that the genes mutated in genetic diseases can display aRMAE. Thus, variable proportions of cells in heterozygous individuals express the different genotypes at the transcript level. We have coined the term transcriptotype to describe this situation, to facilitate its clinical adoption. In autosomal dominant diseases, this effect would manifest as a proportion of cells with only WT transcripts in unaffected or mildly affected individuals.

Thus, in clinical genetics, a patient’s genotype may be WT/MUT, but an empirical and probably mutation-specific determination is required if a biallelic transcriptotype of WT(50)/MUT(50) suffices to cause disease, as expected in fully penetrant dominant diseases, and if a WT(80)/MUT(20) or WT(60)/MUT(40) transcriptotype is sufficient for some expression of the disease. Interestingly, this suggests that genetic diagnosis is not static and that just as we monitor thyroid function every few months in patients with thyroid conditions, so we might need to monitor transcriptional activity in genetic diseases, probably in the cell types in which genotype is most closely linked to phenotype.

This brings us to the questions of where and why aRMAE might have developed, as aRMAE is probably more common than not in any given individual and probably affects hundreds if not thousands of alleles at the same time. What if we introduced the idea that individuals undergo the positive and negative transcriptional selection of specific alleles throughout their lifespan? Unlike classic evolutionary selection, which operates across generations through reproductive success and survival, this personal evolution is driven by intraindividual changes that are not necessarily dependent on transmission to progeny. Allelic variants may gain or lose their transcriptional advantage over time via aRMAE, and these shifts may occur in a cell type– and/or tissue-specific manner, shaped by environmental exposures and physiological context (Fig. 4). In a way, our life experiences (environment, nutrition, physical and mental activity) all contribute to and fine-tune our inherited and acquired genetics via aRMAE as we personally evolve into our best selves, defined as a malleable transcriptotype and not a static genotype.

I would like to thank Dr. Justin Taft and Dr. Eugene Rudensky for thoughtful suggestions.

Funding for this work was provided by National Institutes of Health grants: P01AI186771, AI67802, HD108467, and AI127372 to D.Bogunovic.

Author contributions: Dusan Bogunovic: conceptualization, funding acquisition, methodology, project administration, resources, visualization, and writing—original draft, review, and editing.