
MixMIL integrates GLMM and MIL to enhance single-cell data analysis by modeling cell state heterogeneity efficiently.


coremsg

mixed-models-with-multiple-instance-learning-in-single-cell-data-analysis


Mixed Models with Multiple Instance Learning in Single-Cell Data Analysis


title_rewrite


Caduceus ist das erste RC-äquivariante bidirektionale Langstrecken-DNA-Sprachmodell, das HyenaDNA und Transformer-Modelle in der Genomik übertrifft.


caduceus-bi-directional-equivariant-long-range-dna-sequence-modeling


Caduceus: Bi-Directional Equivariant Long-Range DNA Sequence Modeling



Caduceus introduces innovative bi-directional and reverse complement equivariant modeling for DNA sequences, outperforming existing models on genomics tasks.



DNABERT-2 introduces efficient genome tokenization with Byte Pair Encoding, enhancing model performance for multi-species genomes.


dnabert-2-efficient-multi-species-genome-model-and-benchmark


DNABERT-2: Efficient Multi-Species Genome Model and Benchmark



Integrating scRNA-seq data into GPM enhances understanding of trait variation.


Genotype-phenotype mapping (GPM) or the association of trait variation to genetic variation has been a long-lasting problem in biology. The existing approaches to this problem allowed researchers to partially understand within- and between-species variation as well as the emergence or evolution of phenotypes. However, traditional GPM methods typically ignore the transcriptome or have low statistical power due to challenges related to dataset scale. Thus, it is not clear to what extent selection modulates transcriptomes and whether cis- or trans-regulatory elements are more important. To overcome these challenges, we leveraged the cost efficiency and scalability of single-cell RNA sequencing (scRNA-seq) by collecting data from 18,233 yeast cells from 4,489 segregants of a cross between the laboratory strain BY4741 and the vineyard strain RM11-1a. More precisely, we performed eQTL mapping with the scRNA-seq data to identify single-cell eQTL (sc-eQTL) and transcriptome variation patterns associated to fitness variation inferred from the segregants’ bulk fitness assay. Due to the larger scale of our dataset, we were able to recapitulate results from decades of work in GPM from yeast bulk assays while revealing new associations between phenotypic and transcriptomic variations. The multidimensionality of this dataset also allowed us to measure phenotype and expression heritability and partition the variance of cell fitness into genotype and expression components to highlight selective pressure at both levels. Altogether these results suggest that integrating large-scale scRNA-seq data into GPM improves our understanding of trait variation in the context of transcriptomic regulation.

### Competing Interest Statement

The authors have declared no competing interest.

Refining the resolution of the yeast genotype-phenotype map using single-cell RNA-sequencing

unveiling-yeast-genotype-phenotype-map-with-scrna-seq


Unveiling Yeast Genotype-Phenotype Map with scRNA-seq



Efficiently tokenizing genomes with BPE improves model performance.



Applying reinforcement learning to genome assembly shows promise but faces challenges in real-world scenarios.


reinforcement-learning-approach-to-genome-assembly


Reinforcement Learning Approach to Genome Assembly



DNAの長距離シーケンスモデリングにおける双方向同変性の重要性と効果的なアーキテクチャの提案。



The DDT system plays a crucial role in genome maintenance, preventing large deletions and promoting replication stability.


DNA damage tolerance (DDT) enables replication to continue in the presence of fork stalling lesions. To determine the molecular and genomic impact of a global DDT defect, we studied PcnaK164R/-;Rev1-/- compound mutants. Double mutant (DM) cells displayed increased replication stress, hypersensitivity to genotoxic agents, replication speed, and repriming. A whole genome CRISPR-Cas9 screen revealed a strict reliance of DM cells on the CST complex, where CST promotes fork stability. Whole genome sequencing indicated that this DM DDT defect favors the generation of large, replication-stress inducible deletions of 0.4-4.0kbp, defined as type 3 deletions. Junction break sites of these deletions revealed preferential microhomology preferences of 1-2 base pairs, differing from the smaller type 1 and type 2 deletions. These differential characteristics suggest the existence of molecularly distinct deletion pathways. Type 3 deletions are abundant in human tumors, can dominate the deletion landscape and are associated with DNA damage response status and treatment modality. Our data highlight the essential contribution of the DDT system to genome maintenance and type 3 deletions as mutational signature of replication stress. The unique characteristics of type 3 deletions implicate the existence of a novel deletion pathway in mice and humans that is counteracted by DDT.

### Competing Interest Statement

The authors have declared no competing interest.

Molecular dependencies and genomic consequences of a global DNA damage tolerance defect

molecular-dependencies-and-genomic-consequences-of-a-global-dna-damage-tolerance-defect


Molecular Dependencies and Genomic Consequences of a Global DNA Damage Tolerance Defect



CRISPR-based CTRL-Mutagenesis induces diverse mutations in target regions for functional analysis of non-coding elements.


Functional regions that regulate biological phenomena are interspersed throughout eukaryotic genomes. The most definitive approach for identifying such regions is to confirm the phenotype of cells or organisms in which specific regions have been mutated or removed from the genome. This approach is invaluable for the functional analysis of genes with a defined functional element, the protein-coding sequence. In contrast, no functional analysis platforms have been established for the study of cis-elements or microRNA cluster regions consisting of multiple microRNAs with functional overlap. Whole-genome mutagenesis approaches, such as via N -ethyl- N -nitrosourea (ENU) and gene trapping, have greatly contributed to elucidating the function of coding genes. These methods almost never induce deletions of genomic regions or multiple mutations within a narrow region. In other words, cis-elements and microRNA clusters cannot be effectively targeted in such a manner. Herein, we established a novel region-specific random mutagenesis method named CRISPR-& Transposase-based RegionaL Mutagenesis (CTRL-Mutagenesis). We demonstrated that CTRL-mutagenesis randomly induces diverse mutations within target regions in murine embryonic stem cells. Comparative analysis of mutants harbouring subtly different mutations within the same region would facilitate the further study of cis-element and microRNA clusters.

![Figure][1]</img>

### Competing Interest Statement

The authors have declared no competing interest.

 [1]: pending:yes

Regional random mutagenesis driven by multiple sgRNAs and diverse on-target genome editing events to identify functionally important elements in non-coding regions

regional-random-mutagenesis-for-functional-elements-identification


Regional Random Mutagenesis for Functional Elements Identification
