# README

Application of ChromActivity to IHEC cell types (1698 epigenomes).

## ChromActivity annotations

Batch: `2023-09-06 (final)`

Input data: 
- observed signal tracks (imputed tracks if observed track unavailable) for the following marks: H3K27ac, H3K27me3, H3K36me3, H3K4me1, H3K4me3, H3K9me3
- observed peaks (imputed peaks if observed peaks unavailable) for the following marks: H3K27ac, H3K27me3, H3K36me3, H3K4me1, H3K4me3, H3K9me3
- chromatin state annotations (18 state model)

See the ChromActivity preprint for additional methods:

    Dincer, Tevfik Umut, and Jason Ernst. "Integrative epigenomic and functional characterization assay based annotation of regulatory activity across diverse human cell types." bioRxiv (2023).


### Directory structure:

ChromScore BigWig tracks: 
    
    chromactivity_annotations/ChromScore_bigwigs/_IHEC_IDENTIFIER_.ChromScore.bw


ChromScoreHMM BigBed intervals: 
    
    chromactivity_annotations/ChromScoreHMM_bigbeds/_IHEC_IDENTIFIER_.chromscorehmm.dense.hg38.bb


ChromScoreHMM bed intervals (gzipped): 
    
    chromactivity_annotations/ChromScoreHMM_beds/_IHEC_IDENTIFIER_.chromscorehmm.dense.hg38.bed.gz


Checksums for each file:

    chromactivity_annotations/ChromScore_bigwigs/md5sum.txt
    chromactivity_annotations/ChromScoreHMM_bigbeds/md5sum.txt
    chromactivity_annotations/ChromScoreHMM_beds/md5sum.txt


## ChromActivity serialized models

ChromActivity scikit-learn serialized model:

    models/chromactivity.model

```py
# Python deserialization
import joblib
model = joblib.load("models/chromactivity.model")
```

ChromScoreHMM model (for ChromHMM):

    models/2023-06-17_chromscorehmm.model_15.txt


## Contact information and more 

Contact: dincer@ucla.edu

Code repository: https://github.com/ernstlab/chromactivity

Pre-print:

Dincer, Tevfik Umut, and Jason Ernst. "Integrative epigenomic and functional characterization assay based annotation of regulatory activity across diverse human cell types." bioRxiv (2023).