Predict TF binding regions¶

The predict_tf_binding_regions command uses ChromBERT’s learned co-association patterns to predict TF bindings (e.g., ChIP-seq) for factor–cell pairs where experimental data is unavailable.

Note: The remaining examples show command-line usage only (bash).

For the Python API, see `examples/api/predict_tf_binding_regions.ipynb <../api/predict_tf_binding_regions>`__.

If you need to use Apptainer container, please refer to the `apptainer_use.ipynb <apptainer_use.ipynb>`__ tutorial for detailed instructions on using apptainer exec with chrombert-tools.

For more details, please refer to the `predict_tf_binding_regions <https://chrombert-tools.readthedocs.io/en/latest/commands/predict_tf_binding_regions.html>`__ command documentation.

[1]:

import pandas as pd
import numpy as np
import os

[ ]:

# options parameter
!chrombert-tools predict_tf_binding_regions -h

Usage: chrombert-tools predict_tf_binding_regions [OPTIONS]

  Predict TF binding on specified regions across cell types.

Options:
  --region FILE                   Region BED file.  [required]
  --cistrome TEXT                 factor:cell e.g.
                                  BCL11A:GM12878;BRD4:MCF7;CTCF:HepG2. Use ';'
                                  to separate multiple cistromes.  [required]
  --odir DIRECTORY                Output directory.  [default: ./output]
  --oname TEXT                    Output name prefix.  [default:
                                  cistrome_impute]
  --genome [hg38|mm10]            Genome.  [default: hg38]
  --resolution [1kb]              Resolution. Only supports 1kb resolution in
                                  imputing cistromes task.  [default: 1kb]
  --batch-size INTEGER            Batch size.  [default: 4]
  --num-workers INTEGER           Dataloader workers.  [default: 8]
  --chrombert-cache-dir DIRECTORY
                                  ChromBERT cache directory (containing
                                  config/ and anno/ subfolders).  [default:
                                  ~/.cache/chrombert/data]
  -h, --help                      Show this message and exit.

[ ]:

# --cistrome: # your cistrome: TF:cell_type, separated by semicolons.
# --region: focus regions
# --odir: output directory
# --genome: genome
# --resolution: resolution

!chrombert-tools predict_tf_binding_regions \
    --cistrome "BCL11A:GM12878;BRD4:MCF7;CTCF:HepG2;MYC:H1;MYC:h9;SPI1:GSM2702714" \
    --region "../data/CTCF_ENCFF664UGR_sample100.bed" \
    --odir "./output_predict_tf_binding_regions" \
    --genome "hg38" \
    --resolution "1kb"

Region summary - total: 100, overlapping with ChromBERT: 100 (one region may overlap multiple ChromBERT regions, we keep overlaps with ≥50% coverage of either the ChromBERT bin or the input region), non-overlapping: 0
celltype: h1 has no corresponding wild type dnase data in ChromBERT.
Note: All cistromes names were converted to lowercase for matching.
Cistromes count summary - requested: 6, matched in ChromBERT: 5, not found: 1, not found cistromes: ['myc:h1']
ChromBERT cistromes metas: /mnt/Storage/home/chenqianqian/.cache/chrombert/data/config/hg38_6k_meta.tsv
Your supervised_file does not contain the 'label' column. Please verify whether ground truth column ('label') is required. If it is not needed, you may disregard this message.
Your supervised_file does not contain the 'label' column. Please verify whether ground truth column ('label') is required. If it is not needed, you may disregard this message.
Load pretrained ckpt /mnt/Storage/home/chenqianqian/.cache/chrombert/data/checkpoint/hg38_6k_1kb_pretrain.ckpt successfully!
Loading checkpoint from /mnt/Storage/home/chenqianqian/.cache/chrombert/data/checkpoint/hg38_6k_1kb_prompt_cistrome.ckpt
Loading from pl module, remove prefix 'model.'
Loading from pl module, replace 'pretrain_model' with 'pretrain_model.chrombert'
Loaded 112/112 parameters
Imputing cistromes: 100%|███████████████████████| 25/25 [00:05<00:00,  4.79it/s]

Finished imputing cistromes on specific regions.
Focus region summary - total: 100, overlapping with ChromBERT: 100, non-overlapping: 0
Overlapping regions BED file: ./output_predict_tf_binding_regions/overlap_region.bed
Non-overlapping regions BED file: ./output_predict_tf_binding_regions/no_overlap_region.bed
Results saved to: ./output_predict_tf_binding_regions/results_prob_df.csv
Results track files saved to: ./output_predict_tf_binding_regions/*.bw

[6]:

# results_pro_df: Imputed peak probabilities.
results_pro_df = pd.read_csv("./output_predict_tf_binding_regions/results_prob_df.csv")
results_pro_df

[6]:

	input_chrom	input_start	input_end	chrombert_build_region_index	chrombert_start	chrombert_end	bcl11a:gm12878	brd4:mcf7	ctcf:hepg2	myc:h9	spi1:gsm2702714
0	chr1	37989946	37990368	32658	37990000	37991000	0.781250	0.660156	0.984375	0.972656	0.632812
1	chr11	2400199	2400617	289179	2400000	2401000	0.664062	0.570312	0.972656	0.882812	0.917969
2	chr12	6778809	6779319	391108	6779000	6780000	0.527344	0.412109	0.980469	0.871094	0.503906
3	chr12	52980788	52981316	424926	52981000	52982000	0.174805	0.601562	0.976562	0.812500	0.345703
4	chr12	53676021	53676448	425578	53676000	53677000	0.494141	0.699219	0.968750	0.945312	0.570312
...	...	...	...	...	...	...	...	...	...	...	...
95	chr6	53171843	53172315	1660979	53172000	53173000	0.408203	0.474609	0.988281	0.566406	0.617188
96	chr6	131628105	131628616	1713078	131628000	131629000	0.632812	0.667969	0.988281	0.894531	0.773438
97	chr6	158704189	158704642	1735665	158704000	158705000	0.558594	0.251953	0.972656	0.613281	0.554688
98	chr9	128117589	128118035	2049996	128117000	128118000	0.597656	0.468750	0.972656	0.812500	0.468750
99	chr9	136122853	136123320	2057396	136123000	136124000	0.167969	0.310547	0.968750	0.394531	0.400391

100 rows × 11 columns

[ ]:

[ ]: