def script__summarize_depth(depth_file, summarize_contig_depth_yaml,
summarize_binned_depth_yaml):
depth_file = str(depth_file)
summarize_contig_depth_yaml = str(summarize_contig_depth_yaml)
summarize_binned_depth_yaml = str(summarize_binned_depth_yaml)
config = datahandling.load_config()
"""
Break this into 2?
"""
depth_dict = {}
with open(depth_file, "r") as input_file:
for line in input_file:
if line[0] != "#":
contig = line.split("\t")[0]
depth = int(line.split("\t")[2].strip())
if contig not in depth_dict:
depth_dict[contig] = {}
if depth in depth_dict[contig]:
depth_dict[contig][depth] += 1
else:
depth_dict[contig][depth] = 1
contig_depth_summary_dict = {}
for contig in depth_dict:
total_depth = 0
total_length = 0
for depth in depth_dict[contig]:
length = depth_dict[contig][depth]
total_depth = total_depth + (length * depth)
total_length = total_length + length
contig_depth_summary_dict[contig] = {
"coverage": total_depth / total_length,
"total_depth": total_depth,
"total_length": total_length
}
# Removing as it's also looped in vcf so no need to check twice
# dict is made now to cycle over on range and on contigs
# binned_depth_summary_dict = {}
binned_depth = [0] * 100
# depth_limits = config["serum"]["summarize"]["depth_range"]
# depth_range = list(range(depth_limits[0], depth_limits[1]))
# for bound in depth_range:
# binned_depth_summary_dict[bound] = 0
for contig in depth_dict:
for depth in depth_dict[contig]:
for i in range(1, 100):
if depth >= i:
binned_depth[i - 1] += depth_dict[contig][depth]
datahandling.save_yaml({"contig_depth": contig_depth_summary_dict},
summarize_contig_depth_yaml)
datahandling.save_yaml({"binned_depth": binned_depth},
summarize_binned_depth_yaml)
return 0
import pkg_resources
from ruamel.yaml import YAML
import os
import re
from bifrostlib import datahandling
import sys
config = datahandling.load_config()
global GLOBAL_BIN_VALUES
GLOBAL_BIN_VALUES = [1, 10, 25]
def extract_contigs_sum_cov(file_path, key, data_dict):
yaml = datahandling.load_yaml(file_path)
data_dict["results"][key] = yaml
for bin_value in GLOBAL_BIN_VALUES:
total_length = 0
total_depth = 0
total_contigs = 0
for contig in yaml["contig_depth"]:
if yaml["contig_depth"][contig]["coverage"] >= float(bin_value):
total_length += yaml["contig_depth"][contig]["total_length"]
total_depth += yaml["contig_depth"][contig]["total_depth"]
total_contigs += 1
data_dict["summary"]["bin_contigs_at_{}x".format(
bin_value)] = total_contigs
data_dict["summary"]["bin_length_at_{}x".format(
bin_value)] = total_length
data_dict["summary"]["bin_coverage_at_{}x".format(bin_value)] = float(
total_depth / total_length)