Skip to content

Commit

Permalink
enable user-selected thresholds of contamination and completeness on …
Browse files Browse the repository at this point in the history 
…html page
  • Loading branch information
@fernandomeyer
fernandomeyer committed Jan 3, 2018
1 parent 9adbf4d commit 0da8bf1
Show file tree
Hide file tree
Showing 2 changed files with 54 additions and 105 deletions.
6 changes: 3 additions & 3 deletions index.html
View file

Large diffs are not rendered by default.

153 changes: 51 additions & 102 deletions src/html_plots.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,9 @@

import argparse
import datetime
import re
from math import pi
from collections import OrderedDict

import bokeh.models.widgets.tables
import numpy as np
Expand Down Expand Up @@ -48,85 +50,12 @@
COLORS_10 = palettes.d3['Category10'][10]
HEATMAP_COLORS = list(reversed(palettes.RdYlBu[10])) + ['white']

COMPL_0_5 = '>0.5compl'
COMPL_0_9 = '>0.9compl'
COMPL_0_7 = '>0.7compl'
CONT_0_1 = '<0.1cont'
CONT_0_05 = '<0.05cont'

NO_COLOR_COLUMNS = [labels.SEM_PRECISION, labels.SEM_RECALL, labels.STD_DEV_PRECISION, labels.STD_DEV_RECALL]

CONTAMINATION_COMPLETENESS_COLUMNS = [
'{}{}'.format(COMPL_0_5, CONT_0_1),
'{}{}'.format(COMPL_0_7, CONT_0_1),
'{}{}'.format(COMPL_0_9, CONT_0_1),
'{}{}'.format(COMPL_0_5, CONT_0_05),
'{}{}'.format(COMPL_0_7, CONT_0_05),
'{}{}'.format(COMPL_0_9, CONT_0_05)]

FILE_COMPL_0_5 = '>0.5compl'
FILE_COMPL_0_9 = '>0.9compl'
FILE_COMPL_0_7 = '>0.7compl'
FILE_CONT_0_1 = '<0.1cont'
FILE_CONT_0_05 = '<0.05cont'

# This array is used when the columns of the file are parsed
CONTAMINATION_COMPLETENESS_COLUMNS_FILE = ['{}{}'.format(FILE_COMPL_0_5, FILE_CONT_0_1),
'{}{}'.format(FILE_COMPL_0_7, FILE_CONT_0_1),
'{}{}'.format(FILE_COMPL_0_9, FILE_CONT_0_1),
'{}{}'.format(FILE_COMPL_0_5, FILE_CONT_0_05),
'{}{}'.format(FILE_COMPL_0_7, FILE_CONT_0_05),
'{}{}'.format(FILE_COMPL_0_9, FILE_CONT_0_05)]

COL_DESCRIPTION = "description"
COL_TITLE = "title"
COL_FILE_ID = "file_id"

COMPLETION_COL_0_9 = {
COL_DESCRIPTION: "number of bins with more than 90% completeness",
COL_TITLE: ">90% completeness",
}

COMPLETION_COL_0_7 = {
COL_DESCRIPTION: "number of bins with more than 70% completeness",
COL_TITLE: ">70% completeness",
COL_FILE_ID: '>0.7compl'
}

COMPLETION_COL_0_5 = {
COL_DESCRIPTION: "number of bins with more than 50% completeness",
COL_TITLE: ">50% completeness",
COL_FILE_ID: '>0.5compl'
}

DESCRIPTION_COMPLETENESS_COL = {
COMPL_0_5: COMPLETION_COL_0_5,
FILE_COMPL_0_5: COMPLETION_COL_0_5,
COMPL_0_7: COMPLETION_COL_0_7,
FILE_COMPL_0_7: COMPLETION_COL_0_7,
COMPL_0_9: COMPLETION_COL_0_9,
FILE_COMPL_0_9: COMPLETION_COL_0_9
}

DESCRIPTION_COMPLETENESS_COL_X = {
FILE_COMPL_0_5: COMPLETION_COL_0_5,
FILE_COMPL_0_7: COMPLETION_COL_0_7,
FILE_COMPL_0_9: COMPLETION_COL_0_9
}

DESCRIPTION_CONTAMINATION_ROW = {
CONT_0_1: {
COL_DESCRIPTION: "number of bins with less than 10% contamination",
COL_TITLE: "<10% contamination",
COL_FILE_ID: '<0.1cont'
},
CONT_0_05: {
COL_DESCRIPTION: "number of bins with less than 5% contamination",
COL_TITLE: "<05% contamination",
COL_FILE_ID: '<0.05cont'
}
}

ID_SUMMARY = "summary"
ID_CONTAMINATION_COMPLETENESS = "contamination_completeness"
ID_PRECISION_VS_RECALL_TOOLS = "purity_vs_completeness_tools"
Expand Down Expand Up @@ -200,30 +129,40 @@ def create_title_a(href, number, name):
return [div]


def create_contamination_completeness_table(df):
from collections import OrderedDict
contamination_completenes_row_arr = []
def get_contamination_completeness_thresholds(df):
contamination_completeness_cols = [c for c in df.columns]
completeness_thr = []
contamination_thr = []
for col in contamination_completeness_cols:
values = col.split('<')
completeness_item = values[0]
contamination_item = '<' + values[1]
if completeness_item not in completeness_thr:
completeness_thr.append(completeness_item)
if contamination_item not in contamination_thr:
contamination_thr.append(contamination_item)
return completeness_thr, contamination_thr

for index, row in df.iterrows():
list(filter(lambda x: CONT_0_1 in x, row.index))

def get_row_for_contamination(row, contamination):
def create_contamination_completeness_table(df, completeness_thr, contamination_thr):
contamination_completenes_row_arr = []
for index, row in df.iterrows():
for contamination_item in contamination_thr:
cols = OrderedDict()
cols['Tool'] = "{} {}".format(row.name, DESCRIPTION_CONTAMINATION_ROW[contamination][COL_TITLE])
cols[COMPL_0_5] = row["{}{}".format(COMPL_0_5, contamination)]
cols[COMPL_0_7] = row["{}{}".format(COMPL_0_7, contamination)]
cols[COMPL_0_9] = row["{}{}".format(COMPL_0_9, contamination)]
return cols

contamination_completenes_row_arr.append(get_row_for_contamination(row, CONT_0_1))
contamination_completenes_row_arr.append(get_row_for_contamination(row, CONT_0_05))
contamination_val = format(float(re.match("\d+\.\d+", contamination_item[1:]).group()) * 100, '.0f')
cols['Tool'] = '{} <{}% contamination'.format(row.name, contamination_val)
for completeness_item in completeness_thr:
cols[completeness_item] = row['{}{}'.format(completeness_item, contamination_item)]
contamination_completenes_row_arr.append(cols)

df = pd.DataFrame(contamination_completenes_row_arr)

def create_table_column(field):
if field == "Tool":
return widgets.TableColumn(title=field, field=field, width=600)
else:
return widgets.TableColumn(title=DESCRIPTION_COMPLETENESS_COL[field][COL_TITLE], field=field)
completeness_val = format(float(re.match("\d+\.\d+", field[1:]).group()) * 100, '.0f')
return widgets.TableColumn(title='>{}% completeness'.format(completeness_val), field=field)

dt = DataTable(source=bokeh.models.ColumnDataSource(df),
columns=list(map(lambda x: create_table_column(x), df.columns.values)),
Expand Down Expand Up @@ -415,13 +354,11 @@ def save_html_file(path, elements):

def build_html_summary_path(precision_recall_paths, names, summary, html_output):
summary_df = pd.DataFrame.from_csv(summary, sep='\t', header=0)
summary_df = summary_df.rename(index=str, columns=dict(zip(CONTAMINATION_COMPLETENESS_COLUMNS_FILE,
CONTAMINATION_COMPLETENESS_COLUMNS)))
build_html(precision_recall_paths, names, summary_df, html_output, CONTAMINATION_COMPLETENESS_COLUMNS, NO_COLOR_COLUMNS)
build_html(precision_recall_paths, names, summary_df, html_output, NO_COLOR_COLUMNS)


def build_html(precision_recall_paths, names, summary, html_output,
contamination_completeness_cols=CONTAMINATION_COMPLETENESS_COLUMNS, std_dev_sem_columns=NO_COLOR_COLUMNS):
def build_html(precision_recall_paths, names, summary, html_output, std_dev_sem_columns=NO_COLOR_COLUMNS):
contamination_completeness_cols = [c for c in summary.columns if c.startswith('>')]
element_column = list()

summary.insert(0, "Tool", summary.index)
Expand All @@ -444,7 +381,7 @@ def create_entry(k, v):
return "<li><strong>{0}: </strong>{1}</li>".format(k, v)

html_listing = list(map(lambda k: create_entry(k, labels.abbreviations[k]),
list(filter(lambda k: k not in CONTAMINATION_COMPLETENESS_COLUMNS_FILE,
list(filter(lambda k: k not in contamination_completeness_cols,
labels.abbreviations.keys()))))

element_column.append(
Expand All @@ -453,22 +390,34 @@ def create_entry(k, v):

element_column.append(create_summary_heatmap(df_without_contam_complete, std_dev_sem_columns))

completion_contamination_col = map(lambda k: create_entry(DESCRIPTION_COMPLETENESS_COL_X[k][COL_TITLE],
DESCRIPTION_COMPLETENESS_COL_X[k][COL_DESCRIPTION]),
DESCRIPTION_COMPLETENESS_COL_X.keys())
completeness_thr, contamination_thr = get_contamination_completeness_thresholds(summary[contamination_completeness_cols])

completeness_thr_formatted = []
for completeness_item in completeness_thr:
value = format(float(re.match("\d+\.\d+", completeness_item[1:]).group()) * 100, '.0f')
completeness_thr_formatted.append('{}% completeness'.format(value))

contamination_thr_formatted = []
for contamination_item in contamination_thr:
value = format(float(re.match("\d+\.\d+", contamination_item[1:]).group()) * 100, '.0f')
contamination_thr_formatted.append('{}% contamination'.format(value))

def create_entry_completeness(v):
return "<li><strong>>{0}: </strong>number of bins with more than {1}</li>".format(v, v)

def create_entry_contamination(v):
return "<li><strong><{0}: </strong>number of bins with less than {1}</li>".format(v, v)

completion_contamination_row = map(lambda k:
create_entry(DESCRIPTION_CONTAMINATION_ROW[k][COL_TITLE],
DESCRIPTION_CONTAMINATION_ROW[k][COL_DESCRIPTION]),
DESCRIPTION_CONTAMINATION_ROW.keys())
completion_contamination_col = [create_entry_completeness(v) for v in completeness_thr_formatted]
completion_contamination_row = [create_entry_contamination(v) for v in contamination_thr_formatted]

element_column.append(create_subtitle_div(ID_CONTAMINATION_COMPLETENESS, "Contamination and completeness"))

element_column.append(
create_description("Table columns: <ul>{0}</ul> "
"Table rows: <ul>{1}</ul>".format(" ".join(completion_contamination_col),
" ".join(completion_contamination_row))))
element_column.append(create_contamination_completeness_table(summary[contamination_completeness_cols]))
element_column.append(create_contamination_completeness_table(summary[contamination_completeness_cols], completeness_thr, contamination_thr))

element_column.append(create_subtitle_div(ID_PRECISION_VS_RECALL_TOOLS, "Average purity per bin vs. average completeness per genome"))
element_column.append(create_description(DESCRIPTION_PRECISION_RECALL_TOOLS))
Expand Down

0 comments on commit 0da8bf1

Please sign in to comment.