8. BERT was thoroughly tested emphasizing it's broad scope (tested on large-scale data integration tasks with up to 5000 datasets from simulated and experimental data of different quantification techniques and omic types (proteomics, transcriptomics, metabolomics) as well as other datatypes)

6. BERT is super fast (leverages multi-core and distributed-memory systems for up to 11 × runtime improvement)
7. BERT is available on Bioconductor

3. BERT retains data in incomplete datasets upon integration
4. BERT can integrate datasets with different compositions and conditions (integration of unbalanced datasets using covariates)
5. BERT can integrate datasets with unknown conditions (using references)

BERTs features:

1. BERT can integrate molecular data (Omic data) or other data types (e.g. clinical annotations)
2. BERT tolerates missing data of different types in datasets (e.g. missing at random, missing not at random)

High performance data integration for large-scale analyses of incomplete Omic profiles using Batch-Effect Reduction Trees (BERT) - Nature Communications This study presents BERT, an algorithm for high-performance integration of incomplete omics data with robustness to unequal phenotype distribution. It validates the method on simulated and experimenta...

** High performance #dataintegration for large-scale analyses of incomplete Omic profiles **
Meet BERT (our new algorithm Batch-Effect Reduction Trees) - details in #naturecommunications :
www.nature.com/articles/s41...
#transcriptomics #proteomics #metabolomics #dataanalysis #zmnh #ukeHamburg

Including TE transcription profiles into the molecular characterization of ATRTs might reveal new tumor vulnerabilities leading to novel therapeutic interventions, such as immunotherapy.

--> Differentially transcribed TEs in primary samples are partly reflected in established ATRT-SHH and -MYC cell line models.

--> ATRT-MYC showed broadly reduced transcript levels of LINE1 and ERVL-MaLR subfamilies and displayed significantly less LTR and LINE1 loci with bidirectional promoter activity.

--> We investigate the transcriptional profiles of 1.9M LINE1 and LTR elements that differ across ATRT subtypes in primary human samples.

Atypical teratoid rhabdoid tumors (ATRTs) are very aggressive CNS tumors mainly affecting infants.
Transcriptional activity of transposable elements (TEs), like LINE1s and LTRs, is tightly linked with human cancers as a direct consequence of lifting epigenetic repression over TEs.

Expression of LTR and LINE1 transposable elements defines atypical teratoid/rhabdoid tumor subtypes - Acta Neuropathologica Communications Atypical teratoid rhabdoid tumors (ATRTs) are aggressive central nervous system tumors mainly affecting young children. Extensive molecular characterization based on gene expression and DNA methylatio...

I** Expression of LTR and LINE1 transposable elements defines atypical teratoid/rhabdoid tumor subtypes **
Now out in Acta Neuropathol Commun: actaneurocomms.biomedcentral.com/articles/10....
Special thanks to the Roggenbuck Foundation for funding!
#Neurooncology #Neuropathology #zmnh #ukeHamburg

Many thanks to the @dfg.de for funding this research within the #emmynoether programme!

We provide molecular insights into these rare and severe brain diseases and reveal novel implications of the oncogenic factor LIN28A in extracellular matrix integrity

Alterations of brain layer morphology in these models were mapped to spatial proteome patterns that were acquired with the Nanosecond-infrared laser system

•
In contrast sole stabilisation of CTNNB1 resulted in a distinct developmental brain phenotype similar to Lissencephaly Type 1.

In a nutshell:
•
Co-activation of LIN28A and stabilised CTNNB1 in vivo led to pial disruption and neuronal overmigration resembling a servere human developmental disorder of the brain which is called the Cobblestone Lissencephaly Type 2.

Jelenas work on #spatialproteomics of the developing brain has now been published in Molecular md Cellular Proteomics (www.mcponline.org/article/S153... )

#neurooncology #neurodevelopment #proteomics #Neuropathology #zmnh #ukeHamburg

Finally, hypomethylation at the FGFR3 locus, together with increased FGFR3 protein expression, was observed in 97% of cases, highlighting FGFR3 as a potential future treatment target.

DNA methylation profiles were rather homogenous but global DNA hypomethylation was a feature of tumors associated with higher recurrence risk.

Of note, classic histopathological criteria such as atypia and the cell proliferation index were not reproducible across neuropathologists.
Instead, patient age and treatment strategy were key factors associated with survival outcomes in the cohort.

Outcome-associated factors in a molecularly defined cohort of central neurocytoma - Acta Neuropathologica Central neurocytomas (CN) are intraventricular brain tumors predominantly occurring in young adults. Although prognosis is usually favorable, tumor recurrence is common, particularly following subtota...

***News on Neurocytoma***
Read about outcome-associated factors in a molecularly defined cohort of central neurocytoma
in Acta Neuropathologica:
link.springer.com/article/10.1...

Happy to have contributed to the study!
#neurooncology #Neuropathology
#braintumor

Antonia Gocke representing our group at #SNOPeds2025 !

She shows our data on LIN28A in atypical teratoid rhabdoid tumors and proteome data on a large cohort of #ependymoma. Fantastic job Antonia!
#atrt #braintumor #pediatriccancer #kinderkrebs #proteomics #neurooncology #Neuropathology #zmnh

Direct 3D Sampling of the Embryonic Mouse Head: Layer-wise Nanosecond Infrared Laser (NIRL) Ablation from Scalp to Cortex for Spatially Resolved Proteomics Common workflows in bottom-up proteomics require homogenization of tissue samples to gain access to the biomolecules within the cells. The homogenized tissue samples often contain many different cell ...

She further used the technology to analyze models of lissencephaly. Her results provide a thorough proteome characterization of these rare and severe migration disorders of the brain.
pubs.acs.org/doi/10.1021/...

www.biorxiv.org/content/10.1...

The developing brain cortex is a complex structure relying on a specific spatiotemporal orchestration of signalling molecules and pathways.

To grasp this complexity Jelena Navolic established spatial proteome analyses of the brain applying Nanosecond Infrared Laser Ablation (NIRL).

Dear Jelena, now you successfully defended your PhD ‐ congratulations! We are very happy to have you in the team!

#proteomics #etmr #spatialproteomics #zmnh #ukeHamburg #neurooncology #neurodevelopment #neuropathology

Conversely, methods such as KNNImpute/LLSImpute might be biased to compute their estimates from features (e.g. proteins) with lower average variance - yet, we personally made good experiences with KNN-imputed MS-proteomics data in the past.

Contemporary methods uncertainty-quantification might be an interesting approach here, in addition to methods such as MICE.

Given suitable amounts of training data and sufficient regularization, we would expect NN-based (single-imputation) estimates to converge approximately to the first moments of the distribution without reflecting the heteroskedasticity (cf. bias-variance trade-off).

To the best of our knowledge, the impact of heteroskedasticity has not been thoroughly characterized on quantitative data for the majority of Neighbor/Regression-based and NN-based imputation methods.

Computational Methods for Data Integration and Imputation of Missing Values in Omics Datasets Molecular profiling of different omic-modalities (e.g., DNA methylomics, transcriptomics, proteomics) in biological systems represents the basis for research and clinical decision-making. Measurement...

Ever wondered how different biological datasets can be combined in the best way (e.g. data from #transcriptomics #proteomics #DNAmethylomics)?

Have a look at our #review paper describing challenges of #data integration and available #algorithms and tools.

doi.org/10.1002/pmic...