Machine learning-predicted chromatin organization landscape across pediatric tumors.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches

[1] § Methods › Scoring variants with SuPreMo ↔ scripts/SuPreMo.py, lines 312 to 387 · score 0.72 · reverse complement, CORR scores, contact frequency maps, augmented, Disruption scores, filtered · Report a mismatch
[2] § Results › ML predicts disruptions to 3D genome folding resulting from somatic SVs ↔ scripts/SuPreMo.py, lines 312 to 387 · score 0.61 · maps corresponding, CORR score, contact frequency maps, filter, MSE, sequence · Report a mismatch
[3] § Methods › ABC-modified scoring ↔ scripts/get_Akita_scores_utils.py, lines 603 to 709 · score 0.57 · prediction window, disruption track, disruption score, ALT, weighted, MSE · Report a mismatch

Paper

Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

Python · 876 lines · 40 KB · no license · 2 matches

#!/usr/bin/env python
# SuPreMo: Pipeline for generating mutated sequences for input into predictive models and for scoring variants for disruption to genome folding.
# Written in Python v 3.7.11
'''
usage: SuPreMo [-h] [--sequences SEQUENCES] [--fa FASTA] [--roi ROI] [--roi_scales ROI_SCALES [ROI_SCALES ...]] [--genome {hg19,hg38}]
[--Akita_cell_types {HFF,H1hESC,GM12878,IMR90,HCT116} [{HFF,H1hESC,GM12878,IMR90,HCT116} ...]] [--scores {mse,corr,ssi,scc,ins,di,dec,tri,pca} [{mse,corr,ssi,scc,ins,di,dec,tri,pca} ...]]
[--shift_by SHIFT_WINDOW [SHIFT_WINDOW ...]] [--shifts_file SHIFTS_FILE] [--file OUT_FILE] [--dir OUT_DIR] [--limit SVLEN_LIMIT] [--seq_len SEQ_LEN]
[--revcomp {no_revcomp,add_revcomp,only_revcomp}] [--augment] [--get_seq] [--get_tracks] [--get_maps] [--get_Akita_scores] [--nrows NROWS]
Input file
Pipeline for generating mutated sequences for input into predictive models and for scoring variants for disruption to genome folding.
positional arguments:
Input file Input file with variants. Accepted formats are:
VCF file,
TSV file (SVannot output),
BED file,
TXT file.
Can be gzipped. Coordinates should be 1-based and left open (,], except for SNPs (follow vcf 4.1/4.2 specifications). See custom_perturbations.ipynb for BED or TXT file input specifications.
options:
-h, --help show this help message and exit
--sequences SEQUENCES
Input fasta file with sequences. This file is one outputted by SuPreMo.
(default: None)
--fa FASTA Path to reference genome fasta file. Default: data/{genome}.fa, where genome is hg38 or hg19.
(default: None)
--roi ROI Path to text file with regions of intererst (roi) to use for scoring. Tab-delimited text file with required columns: chrom, start, end or chrom, start1, end1, start2, end2 for paired regions. Alternatively, use 'genes' to weight 2 kb window around protein-coding gene transcription start sites.
(default: None)
--roi_scales ROI_SCALES [ROI_SCALES ...]
Scale by which regions of interest (roi) are upweighted compared to the rest of the map. Default is 10 meaning regions of interest have weights of 10 whereas the rest of the bins have weights of 1.
(default: [10])
--genome {hg19,hg38} Genome to be used: hg19 or hg38. (default: ['hg38'])
--Akita_cell_types {HFF,H1hESC,GM12878,IMR90,HCT116} [{HFF,H1hESC,GM12878,IMR90,HCT116} ...]
Cell types to make Akita predictions for: HFF, H1hESC, GM12878, IMR90, and/or HCT116.
(default: ['HFF'])
--scores {mse,corr,ssi,scc,ins,di,dec,tri,pca} [{mse,corr,ssi,scc,ins,di,dec,tri,pca} ...]
Method(s) used to calculate disruption scores. Use abbreviations as follows:
mse: Mean squared error
corr: Spearman correlation
ssi: Structural similarity index measure
scc: Stratum adjusted correlation coefficient
ins: Insulation
di: Directionality index
dec: Contact decay
tri: Triangle method
pca: Principal component method.
(default: ['mse', 'corr'])
--shift_by SHIFT_WINDOW [SHIFT_WINDOW ...]
Values for shifting prediciton windows inputted as space-separated integers (e.g. -1 0 1). Values outside of range -450000 ≤ x ≤ 450000 will be ignored. Prediction windows at the edge of chromosome arms will only be shifted in the direction that is possible (ex. for window at chrom start, a -1 shift will be treated as a 1 shift since it is not possible to shift left).
(default: [0])
--shifts_file SHIFTS_FILE
Path to file with values to shift each variant by. Required when not all variants are shifted by the same amount. File should be a text file with the same number of rows as the input file and a column which contains a shift value (negative or positive integer) by which to shift the variants in the corresponding row of the input file. (default: None)
--file OUT_FILE Prefix for output files. Saved files will overwrite any existing files.
(default: SuPreMo)
--dir OUT_DIR Output directory. If directory already exists, files will be saved in existing directory. If the same files already exists in that directory, new files will overwrite them. (default: SuPreMo_output)
--limit SVLEN_LIMIT Maximum length of variants to be scored. Filtering out variants that are too big can save time and memory. If not specified, will be set to 2/3 of seq_len.
(default: None)
--seq_len SEQ_LEN Length for sequences to generate. Default value is based on Akita requirement. If non-default value is set, get_Akita_scores must be false.
(default: 1048576)
--revcomp {no_revcomp,add_revcomp,only_revcomp}
Option to use the reverse complement of the sequence:
no_revcomp: no, only use the standard sequence;
add_revcomp: yes, use both the standard sequence and its reverse complement;
only_revcomp: yes, only use the reverse complement of the sequence.
The reverse complement of the sequence is only taken with 0 shift.
(default: ['no_revcomp'])
--augment
Only applicable if --get_Akita_scores is specified. Get the mean and median scores from sequences with specified shifts and reverse complement. If augment is used but shift and revcomp are not specified, the following four sequences will be used:
1) no augmentation: 0 shift and no reverse complement,
2) +1bp shift and no reverse complement,
3) -1bp shift and no reverse complement,
4) 0 shift and take reverse complement.
(default: False)
--get_seq Save sequences for the reference and alternate alleles in fa file format. If --get_seq is not specified, must specify --get_Akita_scores.
Sequence name format: {var_index}_{shift}_{revcomp_annot}_{seq_index}_{var_rel_pos}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken;
seq_index: index for sequences generated for that variant: 0-1 for non-BND reference and alternate sequences and 0-2 for BND left and right reference sequence and alternate sequence;
var_rel_pos: relative position of variant in sequence: list of two for non-BND variant positions in reference and alternate sequence and an integer for BND breakend position in reference and alternate sequences.
There are 2-3 entries per prediction (2 for non-BND variants and 3 for BND variants).
To read fasta file: pysam.Fastafile(filename).fetch(seqname, start, end).upper().
To get sequence names in fasta file: pysam.Fastafile(filename).references.
(default: False)
--get_tracks Save disruption score tracks (448 bins) in npy file format. Only possible for mse and corr scores.
Dictionary item name format: {var_index}_{track}_{shift}_{revcomp_annot}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
track: disruption score track specified;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken.
There is 1 entry per prediction: a 448x1 array.
To read into a dictionary in python: np.load(filename, allow_pickle="TRUE").item()
(default: False)
--get_maps Save predicted contact frequency maps in npy file format.
Dictionary item name format: {var_index}_{shift}_{revcomp_annot}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken.
There is 1 entry per prediction. Each entry contains the following: 2 (3 for chromosomal rearrangements) arrays that correspond to the upper right triangle of the predicted contact frequency maps, the relative variant position in the map, and the first coordinate of the sequence that the map corresponds to.
To read into a dictionary in python: np.load(filename, allow_pickle="TRUE").item()
(default: False)
--get_Akita_scores Get disruption scores. If --get_Akita_scores is not specified, must specify --get_seq. Scores saved in a dataframe with the same number of rows as the input. For multiple alternate alleles, the scores are separated by a comma. To convert the scores from strings to integers, use float(x), after separating rows with multiple alternate alleles. Scores go up to 20 decimal points.
(default: False)
--nrows NROWS Number of rows (perturbations) to read at a time from input. When dealing with large inputs, selecting a subset of rows to read at a time allows scores to be saved in increments and uses less memory. Files with scores and filtered out variants will be temporarily saved in output direcotry. The file names will have a suffix corresponding to the set of nrows (0-based), for example for an input with 2700 rows and with nrows = 1000, there will be 3 sets. At the end of the run, these files will be concatenated into a comprehensive file and the temporary files will be removed.
(default: 1000)
'''
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Parse through arguments
import argparse
class CustomFormatter(argparse.RawTextHelpFormatter, argparse.ArgumentDefaultsHelpFormatter):
pass
parser = argparse.ArgumentParser(
prog = 'SuPreMo',
formatter_class = CustomFormatter,
description='''Pipeline for generating mutated sequences for input into predictive models and for scoring variants for disruption to genome folding.''')
parser.add_argument('in_file',
metavar = 'Input file',
help = ''' Input file with variants. Accepted formats are:
VCF file,
TSV file (SVannot output),
BED file,
TXT file.
Can be gzipped. Coordinates should be 1-based and left open (,], except for SNPs (follow vcf 4.1/4.2 specifications). See custom_perturbations.ipynb for BED or TXT file input specifications.
''',
type = str)
parser.add_argument('--sequences',
dest = 'sequences',
help = ''' Input fasta file with sequences. This file is one outputted by SuPreMo.
''',
type = str,
required = False)
parser.add_argument('--fa',
dest = 'fasta',
help = '''Path to reference genome fasta file. Default: data/{genome}.fa, where genome is hg38 or hg19.
''',
type = str,
required = False)
parser.add_argument('--roi',
dest = 'roi',
help = '''Path to text file with regions of intererst (roi) to use for scoring. Tab-delimited text file with required columns: chrom, start, end or chrom, start1, end1, start2, end2 for paired regions. Alternatively, use 'genes' to weight 2 kb window around protein-coding gene transcription start sites.
''',
type = str,
required = False)
parser.add_argument('--roi_scales',
dest = 'roi_scales',
nargs = '+',
help = '''Scale by which regions of interest (roi) are upweighted compared to the rest of the map. Default is 10 meaning regions of interest have weights of 10 whereas the rest of the bins have weights of 1.
''',
type = int,
default = [10],
required = False)
parser.add_argument('--genome',
dest = 'genome',
nargs = 1,
help = '''Genome to be used: hg19 or hg38.''',
type = str,
choices = ['hg19', 'hg38'],
default = ['hg38'],
required = False)
parser.add_argument('--Akita_cell_types',
dest = 'Akita_cell_types',
nargs = '+',
help = '''Cell types to make Akita predictions for: HFF, H1hESC, GM12878, IMR90, and/or HCT116.
''',
type = str,
choices = ['HFF', 'H1hESC', 'GM12878', 'IMR90', 'HCT116'],
default = ['HFF'],
required = False)
parser.add_argument('--scores',
dest = 'scores',
nargs = '+',
help = '''
Method(s) used to calculate disruption scores. Use abbreviations as follows:
mse: Mean squared error
corr: Spearman correlation
ssi: Structural similarity index measure
scc: Stratum adjusted correlation coefficient
ins: Insulation
di: Directionality index
dec: Contact decay
tri: Triangle method
pca: Principal component method.
''',
type = str,
choices = ['mse', 'corr', 'ssi', 'scc', 'ins', 'di', 'dec', 'tri', 'pca'],
default = ['mse', 'corr'],
required = False)
parser.add_argument('--shift_by',
dest = 'shift_window',
nargs = '+',
help = '''Values for shifting prediciton windows inputted as space-separated integers (e.g. -1 0 1). Values outside of range -450000 ≤ x ≤ 450000 will be ignored. Prediction windows at the edge of chromosome arms will only be shifted in the direction that is possible (ex. for window at chrom start, a -1 shift will be treated as a 1 shift since it is not possible to shift left).
''',
type = int,
default = [0],
required = False)
parser.add_argument('--shifts_file',
dest = 'shifts_file',
help = 'Path to file with values to shift each variant by. Required when not all variants are shifted by the same amount. File should be a text file with the same number of rows as the input file and a column which contains a shift value (negative or positive integer) by which to shift the variants in the corresponding row of the input file.',
type = str,
required = False)
parser.add_argument('--file',
dest = 'out_file',
help = '''Prefix for output files. Saved files will overwrite any existing files.
''',
type = str,
default = 'SuPreMo',
required = False)
parser.add_argument('--dir',
dest = 'out_dir',
help = 'Output directory. If directory already exists, files will be saved in existing directory. If the same files already exists in that directory, new files will overwrite them.',
type = str,
default = 'SuPreMo_output',
required = False)
parser.add_argument('--limit',
dest = 'svlen_limit',
help = '''Maximum length of variants to be scored. Filtering out variants that are too big can save time and memory. If not specified, will be set to 2/3 of seq_len.
''',
type = int,
required = False)
parser.add_argument('--seq_len',
dest = 'seq_len',
help = '''Length for sequences to generate. Default value is based on Akita requirement. If non-default value is set, get_Akita_scores must be false.
''',
type = int,
default = 1048576,
required = False)
parser.add_argument('--revcomp',
dest = 'revcomp',
nargs = 1,
help = '''
Option to use the reverse complement of the sequence:
no_revcomp: no, only use the standard sequence;
add_revcomp: yes, use both the standard sequence and its reverse complement;
only_revcomp: yes, only use the reverse complement of the sequence.
The reverse complement of the sequence is only taken with 0 shift.
''',
type = str,
choices = ['no_revcomp', 'add_revcomp', 'only_revcomp'],
default = ['no_revcomp'],
required = False)
parser.add_argument('--augment',
dest = 'augment',
help = '''
Only applicable if --get_Akita_scores is specified. Get the mean and median scores from sequences with specified shifts and reverse complement. If augment is used but shift and revcomp are not specified, the following four sequences will be used:
1) no augmentation: 0 shift and no reverse complement,
2) +1bp shift and no reverse complement,
3) -1bp shift and no reverse complement,
4) 0 shift and take reverse complement.
''',
action='store_true',
required = False)
parser.add_argument('--get_seq',
dest = 'get_seq',
help = '''Save sequences for the reference and alternate alleles in fa file format. If --get_seq is not specified, must specify --get_Akita_scores.
Sequence name format: {var_index}_{shift}_{revcomp_annot}_{seq_index}_{var_rel_pos}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken;
seq_index: index for sequences generated for that variant: 0-1 for non-BND reference and alternate sequences and 0-2 for BND left and right reference sequence and alternate sequence;
var_rel_pos: relative position of variant in sequence: list of two for non-BND variant positions in reference and alternate sequence and an integer for BND breakend position in reference and alternate sequences.
There are 2-3 entries per prediction (2 for non-BND variants and 3 for BND variants).
To read fasta file: pysam.Fastafile(filename).fetch(seqname, start, end).upper().
To get sequence names in fasta file: pysam.Fastafile(filename).references.
''',
action = 'store_true',
required = False)
parser.add_argument('--get_tracks',
dest = 'get_tracks',
help = '''Save disruption score tracks (448 bins) in npy file format. Only possible for mse and corr scores.
Dictionary item name format: {var_index}_{track}_{shift}_{revcomp_annot}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
track: disruption score track specified;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken.
There is 1 entry per prediction: a 448x1 array.
To read into a dictionary in python: np.load(filename, allow_pickle="TRUE").item()
''',
action='store_true',
required = False)
parser.add_argument('--get_maps',
dest = 'get_maps',
help = '''Save predicted contact frequency maps in npy file format.
Dictionary item name format: {var_index}_{shift}_{revcomp_annot}
var_index: input row number, followed by _0, _1, etc for each allele of variants with multiple alternate alleles;
shift: integer that window is shifted by;
revcomp_annot: present only if reverse complement of sequence was taken.
There is 1 entry per prediction. Each entry contains the following: 2 (3 for chromosomal rearrangements) arrays that correspond to the upper right triangle of the predicted contact frequency maps, the relative variant position in the map, and the first coordinate of the sequence that the map corresponds to.
To read into a dictionary in python: np.load(filename, allow_pickle="TRUE").item()
''',
action='store_true',
required = False)
parser.add_argument('--get_Akita_scores',
dest = 'get_Akita_scores',
help = '''Get disruption scores. If --get_Akita_scores is not specified, must specify --get_seq. Scores saved in a dataframe with the same number of rows as the input. For multiple alternate alleles, the scores are separated by a comma. To convert the scores from strings to integers, use float(x), after separating rows with multiple alternate alleles. Scores go up to 20 decimal points.
''',
action='store_true',
required = False)
parser.add_argument('--nrows',
dest = 'nrows',
help = '''Number of rows (perturbations) to read at a time from input. When dealing with large inputs, selecting a subset of rows to read at a time allows scores to be saved in increments and uses less memory. Files with scores and filtered out variants will be temporarily saved in output direcotry. The file names will have a suffix corresponding to the set of nrows (0-based), for example for an input with 2700 rows and with nrows = 1000, there will be 3 sets. At the end of the run, these files will be concatenated into a comprehensive file and the temporary files will be removed.
''',
type = int,
default = 1000,
required = False)
args = parser.parse_args()
in_file = args.in_file
input_sequences = args.sequences
fasta_path = args.fasta
roi = args.roi
roi_scales = args.roi_scales
genome = args.genome[0]
Akita_cell_types = args.Akita_cell_types
scores_to_use = args.scores
shift_by = args.shift_window
shifts_file = args.shifts_file
out_file = args.out_file
out_dir = args.out_dir
svlen_limit = args.svlen_limit
seq_len = args.seq_len
revcomp = args.revcomp[0]
augment = args.augment
get_seq = args.get_seq
get_tracks = args.get_tracks
get_maps = args.get_maps
get_Akita_scores = args.get_Akita_scores
var_set_size = args.nrows
__version__ = '1.0'
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Adjust inputs from arguments
# Handle paths
import os
from pathlib import Path
# This file path and repo path
repo_path = Path(__file__).parents[1]
# Output directory and file
if not os.path.exists(out_dir):
os.mkdir(out_dir)
out_file = os.path.join(out_dir, out_file)
# Data path
chrom_lengths_path = f'{repo_path}/data/chrom_lengths_{genome}'
centromere_coords_path = f'{repo_path}/data/centromere_coords_{genome}'
# Handle argument dependencies
use_roi = False
if fasta_path is None:
fasta_path = f'{repo_path}/data/{genome}.fa'
if seq_len != 1048576:
get_Akita_scores = False
if svlen_limit is None:
svlen_limit = 2/3*seq_len
elif svlen_limit > 2/3*seq_len:
raise ValueError("Maximum SV length limit should not be >2/3 of sequence length.")
if not get_seq and not get_Akita_scores:
raise ValueError('Either get_seq and/or get_Akita_scores must be specified.')
if roi is not None and not get_Akita_scores:
raise ValueError('get_Akita_scores must be specified to use roi.')
# Adjust shift input: Remove shifts that are outside of allowed range
max_shift = 0.4*seq_len
shift_by = [x for x in shift_by if x > -max_shift and x < max_shift]
# Adjust input for taking the reverse complement
if revcomp == 'no_revcomp':
revcomp_decision = [False]
elif revcomp == 'add_revcomp':
revcomp_decision = [False, True]
elif revcomp == 'only_revcomp':
revcomp_decision = [True]
if augment and shift_by == [0] and revcomp == 'no_revcomp':
shift_by = [-1,0,1]
revcomp_decision = [False, True]
revcomp_decision_i = revcomp_decision
import pysam
if input_sequences is not None:
seq_names = pysam.Fastafile(input_sequences).references
# Create dictionaries to save sequences, maps, and disruption score tracks, if specified
if get_seq:
sequences = {}
if get_maps:
variant_maps = {}
if get_Akita_scores == False:
get_Akita_scores = True
print('Must get scores to get maps. --get_Akita_scores was not specified but will be applied.')
if get_tracks:
variant_tracks = {}
if get_Akita_scores == False:
get_Akita_scores = True
print('Must get scores to get tracks. --get_Akita_scores was not specified but will be applied.')
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Read in (and adjust) data
import pandas as pd
chrom_lengths = pd.read_table(chrom_lengths_path, header = None, names = ['CHROM', 'chrom_max'])
centromere_coords = pd.read_table(centromere_coords_path, sep = '\t')
fasta_open = pysam.Fastafile(fasta_path)
# Assign necessary values to variables across module
# Reading utilities
import sys
sys.path.insert(0, 'scripts/')
import reading_utils
reading_utils.var_set_size = var_set_size
# SuPreMo utilities
import get_seq_utils
get_seq_utils.fasta_open = fasta_open
get_seq_utils.chrom_lengths = chrom_lengths
get_seq_utils.centromere_coords = centromere_coords
get_seq_utils.svlen_limit = svlen_limit
get_seq_utils.seq_length = seq_len
get_seq_utils.half_patch_size = round(seq_len/2)
# SuPreMo-Akita utilities
if get_Akita_scores:
import get_Akita_scores_utils
get_Akita_scores_utils.chrom_lengths = chrom_lengths
get_Akita_scores_utils.centromere_coords = centromere_coords
if roi is None or roi_scales == [0]:
use_roi = False
else:
use_roi = True
import get_roi_utils
get_roi_utils.pred_len = get_Akita_scores_utils.target_length_cropped * get_Akita_scores_utils.bin_size
get_roi_utils.d = get_Akita_scores_utils.bin_size * 5
get_Akita_scores_utils.roi_coords_BED = get_roi_utils.get_roi(roi, genome)
from pybedtools import BedTool
get_Akita_scores_utils.BedTool = BedTool
get_Akita_scores_utils.roi_scales = roi_scales
import sys
import numpy as np
nt = ['A', 'T', 'C', 'G']
var_set = 0
var_set_list = []
print(f'Log file being saved here: {out_file}_log')
while True:
# Read in variants
variants = reading_utils.read_input(in_file, var_set)
if len(variants) == 0:
break
if shifts_file is not None:
variants = pd.concat([variants,
pd.read_csv(shifts_file, sep = '\t', low_memory=False, names = ['shift_by'],
skiprows = 1 + var_set*var_set_size, nrows = var_set_size)],
axis = 1)
# Index input based on row number and create output with same indexes
variants['var_index'] = list(range(var_set*var_set_size, var_set*var_set_size + len(variants)))
variants['var_index'] = variants['var_index'].astype(str)
# If there are multiple alternate alleles, split those into new rows and indexes
if any([',' in x for x in variants.ALT]):
variants = (variants
.set_index(['CHROM', 'POS', 'REF', 'var_index'])
.apply(lambda x: x.str.split(',').explode())
.reset_index())
g = variants.groupby(['var_index'])
variants.loc[g['var_index'].transform('size').gt(1),
'var_index'] += '-'+g.cumcount().astype(str)
variant_scores = pd.DataFrame({'var_index':variants.var_index})
if use_roi:
variant_roi_ids = pd.DataFrame({'var_index':variants.var_index})
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Filter out variants that cannot be scored
# Get indexes for variants to exclude
# Exclude mitochondrial variants
chrM_var = pd.DataFrame({'var_index' : list(variants[variants.CHROM == 'chrM'].var_index),
'reason' : ' Mitochondrial chromosome.'})
# Exclude variants larger than limit
if 'SVLEN' in variants.columns:
too_long_var = pd.DataFrame({'var_index' : [y for x,y in zip(variants.SVLEN, variants.var_index)
if not pd.isnull(x) and abs(int(x)) > svlen_limit],
'reason' : f' SV longer than {svlen_limit}.'})
unsuitable_var = pd.DataFrame({'var_index' : [y for x,y,z in zip(variants.SVTYPE, variants.var_index, variants.ALT)
if not pd.isnull(x) and
x not in ["DEL", "DUP", "INV", "BND"] and
all([g not in nt for g in z])],
'reason' : ' SV type not compatible.'})
else:
too_long_var = pd.DataFrame()
unsuitable_var = pd.DataFrame()
filtered_out = pd.concat([chrM_var, too_long_var], axis = 0)
filtered_out = pd.concat([filtered_out, unsuitable_var], axis = 0)
filtered_out.var_index = filtered_out.var_index.astype('str')
# Save filtered out variants into file
filtered_out.to_csv(f'{out_file}_filtered_out_{var_set}', sep = ':', index = False, header = False)
# Exclude
variants = variants[[x not in filtered_out.var_index.values for x in variants.var_index]]
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Run: Make Akita predictions and calculate disruption scores
# Create log file to save standard output with error messages
std_output = sys.stdout
log_file = open(f'{out_file}_log_{var_set}','w')
sys.stdout = log_file
# Loop through each row (not index) and get disruption scores
for i in range(len(variants)):
variant = variants.iloc[i]
var_index = variant.var_index
CHR = variant.CHROM
POS = variant.POS
REF = variant.REF
ALT = variant.ALT
if 'SVTYPE' in variants.columns:
END = variant.END
SVTYPE = variant.SVTYPE
SVLEN = variant.SVLEN
else:
END = np.nan
SVTYPE = np.nan
SVLEN = 0
if shifts_file is not None:
shift_by = [int(variant.shift_by)]
for shift in shift_by:
# Take reverse complement only with 0 shift
if shift != 0 & True in revcomp_decision:
revcomp_decision_i = [False]
else:
revcomp_decision_i = revcomp_decision
for revcomp in revcomp_decision_i:
try:
if revcomp:
revcomp_annot = '_revcomp'
else:
revcomp_annot = ''
if input_sequences is not None:
# Generate sequences_i from sequence input
if revcomp_annot == '':
sequence_names = [x for x in seq_names if x.startswith(f'{var_index}_{shift}') and
'revcomp' not in x]
elif revcomp_annot == '_revcomp':
sequence_names = [x for x in seq_names if x.startswith(f'{var_index}_{shift}{revcomp_annot}')]
sequences_i = []
for sequence_name in sequence_names:
sequences_i.append(pysam.Fastafile(input_sequences).fetch(sequence_name, 0, seq_len).upper())
sequences_i.append([int(x) for x in sequence_name.split('[')[1].split(']')[0].split('_')])
else:
# Create sequences_i from variant input
sequences_i = get_seq_utils.get_sequences_SV(CHR, POS, REF, ALT, END, SVTYPE, shift, revcomp)
if get_seq:
# Get relative position of variant in sequence
var_rel_pos = str(sequences_i[-1]).replace(', ', '_')
for ii in range(len(sequences_i[:-1][:3])):
sequences[f'{var_index}_{shift}{revcomp_annot}_{ii}_{var_rel_pos}'] = sequences_i[:-1][ii]
if get_Akita_scores:
scores = get_Akita_scores_utils.get_scores(CHR, POS, SVTYPE, SVLEN,
sequences_i, scores_to_use,
shift, revcomp,
get_tracks, get_maps, use_roi, Akita_cell_types)
if get_tracks:
for track in [x for x in scores.keys() if 'track' in x]:
variant_tracks[f'{var_index}_{track}_{shift}{revcomp_annot}'] = scores[track]
del scores[track]
if get_maps:
for map in [x for x in scores.keys() if 'map' in x]:
variant_maps[f'{var_index}_{map}_{shift}{revcomp_annot}'] = scores[map]
del scores[map]
if shifts_file is not None:
shift = 'shifted'
if use_roi:
for roi_ids in [x for x in scores.keys() if 'roi_id' in x]:
variant_roi_ids.loc[variant_scores.var_index == var_index, f'roi_ids_{shift}'] = scores[roi_ids]
del scores[roi_ids]
for score in scores:
variant_scores.loc[variant_scores.var_index == var_index,
f'{score}_{shift}{revcomp_annot}'] = scores[score]
print(str(var_index) + ' (' + str(shift) + f' shift{revcomp_annot})')
except Exception as e:
print(str(var_index) + ' (' + str(shift) + f' shift{revcomp_annot})' + ': Error:', e)
pass
# Write standard output with error messages and warnings to log file
sys.stdout = std_output
log_file.close()
# Combine results from all sets
# Write sequences to fasta file
if get_seq:
if var_set == 0:
sequences_all = sequences.copy()
else:
sequences_all.update(sequences)
# Write scores to data frame
if get_Akita_scores:
# Take average of augmented sequences
if augment:
for score in scores:
cols = [x for x in variant_scores.columns if score in x]
variant_scores[f'{score}_mean'] = variant_scores[cols].mean(axis = 1)
variant_scores[f'{score}_median'] = variant_scores[cols].median(axis = 1)
variant_scores.drop(cols, axis = 1, inplace = True)
# Convert scores from float to string so you can merge scores for variants with multiple alleles
for col in variant_scores.iloc[:,1:].columns:
variant_scores[col] = [format(x, '.20f') for x in variant_scores[col]]
# Join scores for alternate alleles, separated by a comma
if any(['-' in x for x in variant_scores.var_index]):
variant_scores['var_index'] = variant_scores.var_index.str.split('-').str[0]
variant_scores = (variant_scores
.set_index(['var_index'], drop = False)
.rename(columns = {'var_index':'var_index2'})
.groupby('var_index2')
.transform(','.join)
.reset_index()
.drop_duplicates())
if var_set == 0:
variant_scores.to_csv(f'{out_file}_scores_{var_set}', sep = '\t', index = False)
if use_roi:
variant_roi_ids.to_csv(f'{out_file}_roi_ids_{var_set}', sep = '\t', index = False)
else:
variant_scores.to_csv(f'{out_file}_scores_{var_set}', sep = '\t', index = False, header = False)
if use_roi:
variant_roi_ids.to_csv(f'{out_file}_roi_ids_{var_set}', sep = '\t', index = False, header = False)
if get_tracks:
if var_set == 0:
variant_tracks_all = variant_tracks.copy()
else:
variant_tracks_all.update(variant_tracks)
if get_maps:
if var_set == 0:
variant_maps_all = variant_maps.copy()
else:
variant_maps_all.update(variant_maps)
var_set_list.append(var_set)
var_set += 1
# # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Save results
# Write sequences to fasta file
if get_seq:
sequences_fasta = open(f'{out_file}_sequences.fa','w')
for seq_name, sequence in sequences_all.items():
seq_name_line = ">" + seq_name + "\n"
sequences_fasta.write(seq_name_line)
sequence_line = sequence + "\n"
sequences_fasta.write(sequence_line)
sequences_fasta.close()
# Write disruption tracks and/or predictions to array
if get_tracks:
np.save(f'{out_file}_tracks.npy', variant_tracks_all)
if get_maps:
np.save(f'{out_file}_maps.npy', variant_maps_all)
# Combine subset files into one
os.system(f'rm -f {out_file}_filtered_out; \
for file in {out_file}_filtered_out_*; \
do cat "$file" >> {out_file}_filtered_out && rm "$file"; \
done')
os.system(f'rm -f {out_file}_log; \
for file in {out_file}_log_*; \
do cat "$file" >> {out_file}_log && rm "$file"; \
done')
if get_Akita_scores:
os.system(f'rm -f {out_file}_scores; \
for file in {out_file}_scores_*; \
do cat "$file" >> {out_file}_scores && rm "$file"; \
done')
if use_roi:
os.system(f'rm -f {out_file}_roi_ids; \
for file in {out_file}_roi_ids_*; \
do cat "$file" >> {out_file}_roi_ids && rm "$file"; \
done')
# Adjust log file to only have 1 row per variant
if os.path.exists(f'{out_file}_log'):
log_file = pd.read_csv(f'{out_file}_log', names = ['output'], sep = '\t')
# Move warnings (printed 1 line before variant) to variant line
indexes = np.array([[index, index+1] for (index, item) in enumerate(log_file.output) if item.startswith('Warning')])
if len(indexes) != 0:
log_file.loc[indexes[:,1],'output'] = [x+': '+y for x,y in zip(list(log_file.loc[indexes[:,1],'output']),
list(log_file.loc[indexes[:,0],'output']))]
log_file.drop(indexes[:,0], axis = 0, inplace = True)
log_file.to_csv(f'{out_file}_log', sep = '\t', header = None, index = False)

SuPreMo.py at commit 9799281, no license · at the source

Overview

Authors: Ketrin Gjoni^1,2, Shu Zhang^1,2, Rachel E Yan³, Bo Zhang⁴, Daniel Miller⁴, Jeffrey P Greenfield³, Adam Resnick⁴, Nadia Dahmane³, Katherine S Pollard^1,2,5,6

Gladstone Institute of Data Science and Biotechnology, San Francisco, CA 94158 USA
Department of Epidemiology and Biostatistics, University of California, San Francisco, CA 94158 USA
Department of Neurological Surgery, Weill Cornell Medicine, New York, NY 10065 USA
Center for Data-Driven Discovery in Biomedicine, Children’s Hospital of Philadelphia, Philadelphia, PA 19104 USA
Bakar Computational Health Sciences Institute, University of California, San Francisco, CA 94143 USA
Chan Zuckerberg Biohub, San Francisco, CA 94158 USA

Institutions: Gladstone Institutes (United States); University of California, San Francisco (United States); Weill Cornell Medicine (United States); Children's Hospital of Philadelphia (United States); Chan Zuckerberg Biohub San Francisco (United States)

Journal: Scientific reports, volume 16, issue 1, article 10790

Dates: received 29 March 2025; accepted 16 March 2026; published online 28 March 2026

Type: Research article · Language: English

License: CC BY

Identifiers: DOI 10.1038/s41598-026-44925-3 · PMID 41904260 · PMCID PMC13039956 · OpenAlex W7142297467

Open access: gold, a free copy (OpenAlex)

Status: code verified (how the code is found and verified)

Categories: genetics / omics (modality), human (organism), other condition (population)

Methods: Spectral & time-frequency, Statistics, Smoothing, state filtering, decompositions, Connectivity

Keywords: Cancer genomics, Computational models, Machine learning, Epigenomics, Structural variation

MeSH: Chromatin*, Genomic Structural Variation*, Machine Learning*, Neoplasms*, Child, Convolutional Neural Networks, Humans (* major topic)

Topic: Genomics and Chromatin Dynamics (Molecular Biology, Biochemistry, Genetics and Molecular Biology), according to OpenAlex

Funding: Patrick Bayly Marsano Foundation; Ty Louis Campbell Foundation; NIH HHS (R03-DE032505); Gladstone Institutes; Biswas Family Foundation; National Science Foundation; National Institutes of Health (R03-DE032505)

Citations: cited by 2 papers (Europe PMC); 45 references in the paper

Abstract

Structural variants (SVs) are increasingly recognized as important contributors to oncogenesis through their effects on 3D genome folding. Recent advances in whole-genome sequencing have enabled large-scale profiling of SVs across diverse tumors, yet experimental characterization of their individual impact on genome folding remains infeasible. Here, we leveraged a convolutional neural network, Akita, to predict disruptions in genome folding caused by somatic SVs identified in 61 tumor types from the Children’s Brain Tumor Network dataset. Our analysis reveals significant variability in SV-induced disruptions across tumor types, with the most disruptive SVs coming from lymphomas and sarcomas, metastatic tumors, and germline cell tumors. Dimensionality reduction of disruption scores identified five recurrently disrupted regions enriched for high-impact SVs across multiple tumors. Some of these regions are highly disrupted despite not being highly mutated, and harbor tumor-associated genes and transcriptional regulators. To further interpret the functional relevance of high-scoring SVs, we integrated epigenetic data and developed a modified Activity-by-Contact scoring approach to prioritize SVs with disrupted genome contacts at active enhancers. This method highlighted highly disruptive SVs near key oncogenes, as well as novel candidate loci potentially implicated in tumorigenesis. These findings highlight the utility of machine learning for identifying novel SVs, loci, and genetic mechanisms contributing to pediatric cancers. This framework provides a foundation for future studies linking SV-driven regulatory changes to cancer pathogenesis.

Supplementary Information: The online version contains supplementary material available at 10.1038/s41598-026-44925-3.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repositories

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

ENCODE-DCC/chip-seq-pipeline2

License: MIT

State: the link answers, verified on 6 November 2026

Evidence: files inventoried

Commit: 26eeda81a0540dc793fc69b0c390d232ca7ca50a, 15 February 2024

Languages: Python (46), Shell (9)

Size: 225 files, 55 scripts

Software Heritage: not archived

Found in: the text, “Epigenetic data processing”

Holds: README, license file, environment (scripts/requirements.macs2.txt, scripts/requirements.spp.txt, scripts/requirements.txt, dev/docker_image/Dockerfile), tests, continuous integration, documentation

Not found: CITATION.cff

Tools: Matplotlib (6 files), NumPy (6 files), pandas (3 files), SAMtools (2 files), SciPy (2 files), BEDTools (1 file)

Availability: 2 checks, the latest on 6 November 2026: the link answers

6 November 2026: the link answers
30 September 2026: the link answers

57 files

dev/build_on_dx_dockerhub.sh, Shell, 76 lines
dev/test/run_cromwell_server_on_gc.sh, Shell, 15 lines
dev/test/test_py/__init__.py, Python, 1 line
dev/test/test_py/test_encode_lib_genomic.py, Python, 368 lines
dev/test/test_workflow/ref_output/sync.sh, Shell, 3 lines
scripts/build_genome_data.sh, Shell, 351 lines
scripts/download_genome_data.sh, Shell, 297 lines
scripts/install_conda_env.sh, Shell, 85 lines
scripts/uninstall_conda_env.sh, Shell, 11 lines
scripts/update_conda_env.sh, Shell, 20 lines
src/assign_multimappers.py, Python, 73 lines
src/detect_adapter.py, Python, 84 lines
src/dev_check_sync_atac.sh, Shell, 21 lines
src/encode_lib_blacklist_filter.py, Python, 117 lines
src/encode_lib_common.py, Python, 399 lines
src/encode_lib_frip.py, Python, 114 lines
src/encode_lib_genomic.py, Python, 767 lines
src/encode_lib_log_parser.py, Python, 660 lines
src/encode_lib_qc_category.py, Python, 259 lines
src/encode_task_annot_enrich.py, Python, 109 lines
src/encode_task_bam2ta.py, Python, 162 lines
src/encode_task_bam_to_pbam.py, Python, 64 lines
src/encode_task_bowtie2.py, Python, 192 lines
src/encode_task_bwa.py, Python, 280 lines
src/encode_task_choose_ctl.py, Python, 181 lines
src/encode_task_compare_signal_to_roadmap.py, Python, 137 lines
src/encode_task_count_signal_track.py, Python, 115 lines
src/encode_task_filter.py, Python, 438 lines
src/encode_task_frac_mito.py, Python, 86 lines
src/encode_task_fraglen_stat_pe.py, Python, 244 lines
src/encode_task_gc_bias.py, Python, 143 lines
src/encode_task_idr.py, Python, 213 lines
src/encode_task_jsd.py, Python, 133 lines
src/encode_task_macs2_atac.py, Python, 139 lines
src/encode_task_macs2_chip.py, Python, 172 lines
src/encode_task_macs2_signal_track_atac.py, Python, 190 lines
src/encode_task_macs2_signal_track_chip.py, Python, 215 lines
src/encode_task_merge_fastq.py, Python, 105 lines
src/encode_task_overlap.py, Python, 178 lines
src/encode_task_pool_ta.py, Python, 76 lines
src/encode_task_post_align.py, Python, 92 lines
src/encode_task_post_call_peak_atac.py, Python, 104 lines
src/encode_task_post_call_peak_chip.py, Python, 107 lines
src/encode_task_preseq.py, Python, 174 lines
src/encode_task_qc_report.py, Python, 941 lines
src/encode_task_reproducibility.py, Python, 204 lines
src/encode_task_spp.py, Python, 133 lines
src/encode_task_spr.py, Python, 176 lines
src/encode_task_subsample_ctl.py, Python, 66 lines
src/encode_task_trim_adapter.py, Python, 214 lines
src/encode_task_trim_fastq.py, Python, 73 lines
src/encode_task_trimmomatic.py, Python, 204 lines
src/encode_task_tss_enrich.py, Python, 178 lines
src/encode_task_xcor.py, Python, 156 lines
src/trimfastq.py, Python, 255 lines
LICENSE, License, 21 lines
README.md, Text, 180 lines

At the source: github.com/ENCODE-DCC/chip-seq-pipeline2

ENCODE-DCC/atac-seq-pipeline

License: MIT

State: the link answers, verified on 28 October 2026

Evidence: files inventoried

Commit: 47ba8dff9c332e24b48e767303e9fcac98589cf2, 15 February 2024

Languages: Python (45), Shell (9)

Size: 204 files, 54 scripts

Software Heritage: not archived

Found in: the text, “Epigenetic data processing”

Holds: README, license file, environment (scripts/requirements.macs2.txt, scripts/requirements.python2.txt, scripts/requirements.spp.txt, scripts/requirements.txt, dev/docker_image/Dockerfile), tests, continuous integration, documentation

Not found: CITATION.cff

Tools: Matplotlib (6 files), NumPy (6 files), pandas (3 files), SAMtools (2 files), SciPy (2 files), BEDTools (1 file)

Availability: 2 checks, the latest on 28 October 2026: the link answers

28 October 2026: the link answers
27 September 2026: the link answers

56 files

dev/build_on_dx_dockerhub.sh, Shell, 60 lines
dev/test/run_cromwell_server_on_gc.sh, Shell, 15 lines
dev/test/test_py/__init__.py, Python, 1 line
dev/test/test_workflow/ref_output/sync.sh, Shell, 3 lines
scripts/build_genome_data.sh, Shell, 351 lines
scripts/download_genome_data.sh, Shell, 297 lines
scripts/install_conda_env.sh, Shell, 88 lines
scripts/uninstall_conda_env.sh, Shell, 12 lines
scripts/update_conda_env.sh, Shell, 21 lines
src/assign_multimappers.py, Python, 73 lines
src/detect_adapter.py, Python, 84 lines
src/dev_check_sync_atac.sh, Shell, 21 lines
src/encode_lib_blacklist_filter.py, Python, 117 lines
src/encode_lib_common.py, Python, 399 lines
src/encode_lib_frip.py, Python, 114 lines
src/encode_lib_genomic.py, Python, 767 lines
src/encode_lib_log_parser.py, Python, 660 lines
src/encode_lib_qc_category.py, Python, 259 lines
src/encode_task_annot_enrich.py, Python, 109 lines
src/encode_task_bam2ta.py, Python, 163 lines
src/encode_task_bam_to_pbam.py, Python, 64 lines
src/encode_task_bowtie2.py, Python, 192 lines
src/encode_task_bwa.py, Python, 280 lines
src/encode_task_choose_ctl.py, Python, 181 lines
src/encode_task_compare_signal_to_roadmap.py, Python, 137 lines
src/encode_task_count_signal_track.py, Python, 115 lines
src/encode_task_filter.py, Python, 438 lines
src/encode_task_frac_mito.py, Python, 86 lines
src/encode_task_fraglen_stat_pe.py, Python, 244 lines
src/encode_task_gc_bias.py, Python, 143 lines
src/encode_task_idr.py, Python, 213 lines
src/encode_task_jsd.py, Python, 133 lines
src/encode_task_macs2_atac.py, Python, 139 lines
src/encode_task_macs2_chip.py, Python, 172 lines
src/encode_task_macs2_signal_track_atac.py, Python, 190 lines
src/encode_task_macs2_signal_track_chip.py, Python, 215 lines
src/encode_task_merge_fastq.py, Python, 105 lines
src/encode_task_overlap.py, Python, 178 lines
src/encode_task_pool_ta.py, Python, 76 lines
src/encode_task_post_align.py, Python, 92 lines
src/encode_task_post_call_peak_atac.py, Python, 104 lines
src/encode_task_post_call_peak_chip.py, Python, 107 lines
src/encode_task_preseq.py, Python, 174 lines
src/encode_task_qc_report.py, Python, 941 lines
src/encode_task_reproducibility.py, Python, 204 lines
src/encode_task_spp.py, Python, 133 lines
src/encode_task_spr.py, Python, 176 lines
src/encode_task_subsample_ctl.py, Python, 66 lines
src/encode_task_trim_adapter.py, Python, 214 lines
src/encode_task_trim_fastq.py, Python, 73 lines
src/encode_task_trimmomatic.py, Python, 204 lines
src/encode_task_tss_enrich.py, Python, 178 lines
src/encode_task_xcor.py, Python, 156 lines
src/trimfastq.py, Python, 255 lines
LICENSE, License, 21 lines
README.md, Text, 184 lines

At the source: github.com/ENCODE-DCC/atac-seq-pipeline

ketringjoni/SuPreMo

License: none: the authors keep all their rights

State: the link answers, verified on 26 September 2026

Evidence: files inventoried

Commit: 9799281ec9b4ea5ea03702931e57950903f75424, 14 November 2024

Languages: Python (10), Jupyter (5)

Size: 67 files, 15 scripts

Software Heritage: not archived

Found in: “Code availability”

Holds: README, 5 notebooks

Not found: license file, CITATION.cff, environment file, tests, continuous integration, documentation

Tools: pandas (14 files), NumPy (13 files), pysam (6 files), Matplotlib (5 files), SciPy (5 files), BEDTools (4 files), Biopython (3 files), scikit-learn (2 files), TensorFlow (2 files), scikit-image (1 file)

Availability: 1 check, the latest on 26 September 2026: the link answers

26 September 2026: the link answers

16 files

scripts/SuPreMo.py, Python, 876 lines, 2 matches
scripts/get_Akita_scores_utils.py, Python, 718 lines, 1 match
scripts/get_roi_utils.py, Python, 222 lines
scripts/get_seq_utils.py, Python, 1,033 lines
scripts/hicrep.py, Python, 84 lines
scripts/plotting_utils.py, Python, 691 lines
scripts/reading_utils.py, Python, 184 lines
scripts/scoring.py, Python, 746 lines
scripts/test_install_SuPreMo-Akita.py, Python, 41 lines
scripts/test_install_SuPreMo.py, Python, 16 lines
walkthroughs/SuPreMo-Akita_walkthrough.ipynb, Jupyter, 343 lines
walkthroughs/SuPreMo-Akita_weighted_scores.ipynb, Jupyter, 354 lines
walkthroughs/SuPreMo_walkthrough.ipynb, Jupyter, 358 lines
walkthroughs/custom_perturbations.ipynb, Jupyter, 231 lines
walkthroughs/example_application.ipynb, Jupyter, 396 lines
README.md, Text, 290 lines

At the source: github.com/ketringjoni/SuPreMo

Code availability

The code to run SuPreMo-Akita was previously published and can be found at https://github.com/ketringjoni/SuPreMo.

Reproduced under the paper's license (CC BY), from the paper cited above.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

3 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
124 scripts, each with its path and the digest of its content;
3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

geo:GSE162976, at NCBI GEO; found in “Data availability”

Data Availability Statement

ChIP-seq for DIPG007 and KNS42 cell lines were obtained from the Gene Expression Omnibus with accession code GSE162976 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE162976). ChIP-seq for D283 and BT16 were obtained from GSE92585 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92585) and GSE174446 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE174446), respectively. ATAC-seq for D283, DIPG007, and KNS42 cell lines were obtained from GSE92585 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92585), GSE229453 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE229453) and GSE162831 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE162831), respectively. ATAC-seq for BT16 were obtained from 45 (https://paperpile.com/c/cgnLqO/yTPjQ). SuPremo-Akita scores and other results are available with this paper and its Supplementary Information.

The code to run SuPreMo-Akita was previously published and can be found at https://github.com/ketringjoni/SuPreMo.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 30 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 9 authors, 5 keywords, 7 MeSH terms, 7 funders, 44 references.

Cite

This paper

Gjoni, K., Zhang, S., Yan, R. E., Zhang, B., Miller, D., Greenfield, J. P., Resnick, A., Dahmane, N., & Pollard, K. S. (2026). Machine learning-predicted chromatin organization landscape across pediatric tumors. Scientific reports, 16(1), 10790. https://doi.org/10.1038/s41598-026-44925-3

BibTeX

@article{gjoni2026machine, author = {Gjoni, Ketrin and Zhang, Shu and Yan, Rachel E and Zhang, Bo and Miller, Daniel and Greenfield, Jeffrey P and Resnick, Adam and Dahmane, Nadia and Pollard, Katherine S}, title = {{Machine learning-predicted chromatin organization landscape across pediatric tumors}}, journal = {Scientific reports}, year = {2026}, month = mar, volume = {16}, number = {1}, pages = {10790}, publisher = {Nature Publishing Group}, issn = {2045-2322}, doi = {10.1038/s41598-026-44925-3}, url = {https://doi.org/10.1038/s41598-026-44925-3}, pmid = {41904260}, pmcid = {PMC13039956} }

RIS

TY - JOUR AU - Gjoni, Ketrin AU - Zhang, Shu AU - Yan, Rachel E AU - Zhang, Bo AU - Miller, Daniel AU - Greenfield, Jeffrey P AU - Resnick, Adam AU - Dahmane, Nadia AU - Pollard, Katherine S TI - Machine learning-predicted chromatin organization landscape across pediatric tumors T2 - Scientific reports J2 - Sci Rep PY - 2026 DA - 2026/03/28 VL - 16 IS - 1 SP - 10790 SN - 2045-2322 PB - Nature Publishing Group DO - 10.1038/s41598-026-44925-3 UR - https://doi.org/10.1038/s41598-026-44925-3 LA - en ER -

CSL-JSON

{ "id": "10.1038/s41598-026-44925-3", "type": "article-journal", "title": "Machine learning-predicted chromatin organization landscape across pediatric tumors", "container-title": "Scientific reports", "author": [ { "family": "Gjoni", "given": "Ketrin" }, { "family": "Zhang", "given": "Shu" }, { "family": "Yan", "given": "Rachel E" }, { "family": "Zhang", "given": "Bo" }, { "family": "Miller", "given": "Daniel" }, { "family": "Greenfield", "given": "Jeffrey P" }, { "family": "Resnick", "given": "Adam" }, { "family": "Dahmane", "given": "Nadia" }, { "family": "Pollard", "given": "Katherine S" } ], "container-title-short": "Sci Rep", "volume": "16", "issue": "1", "page": "10790", "DOI": "10.1038/s41598-026-44925-3", "PMID": "41904260", "PMCID": "PMC13039956", "ISSN": "2045-2322", "publisher": "Nature Publishing Group", "URL": "https://doi.org/10.1038/s41598-026-44925-3", "language": "en", "issued": { "date-parts": [ [ 2026, 3, 28 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Add the badge to the code's README

If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.

Markdown

<!-- oscr-badge:start -->
[![OSCR](https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3.svg)](https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3)
<!-- oscr-badge:end -->

reStructuredText

.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3.svg
   :target: https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3
   :alt: OSCR

.. oscr-badge:end

HTML

<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3"><img src="https://openscicode.org/badge/doi/10.1038/s41598-026-44925-3.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->

On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".

If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Report a mismatch between the code and the paper · Report a code error (signed in)

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Report a reproduction failure (signed in)

Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.

Machine learning-predicted chromatin organization landscape across pediatric tumors.

Code ↔ Paper

Paper

The authors' code

Overview

Abstract

Repositories

ENCODE-DCC/chip-seq-pipeline2

ENCODE-DCC/atac-seq-pipeline

ketringjoni/SuPreMo

Code availability

Tracing map

Data

Datasets cited

Data Availability Statement

Versions

Version 1, 30 September 2026: the first record

Cite

This paper

Similar papers

Contribute

Claim this paper

Correct its record

Validate its tracing map

Add the badge to the code's README

Request its removal

Discussion, reproductions, activity