u8sand

import pandas as pd

def merge(left, *rights, **kwargs):
  ''' Helper function for many trivial (index based) joins
  Usage:
    merge(frame1, frame2, frame3, how='inner')
  '''
  merged = left
  for right in rights:
    merged = pd.merge(left=merged, left_index=True, right=right, right_index=True, **kwargs)

u8sand

import pandas as pd
import numpy as np

def quantileNormalize_np(mat):
  # sort vector in np (reuse in np)
  sorted_vec = np.sort(mat, axis=0)
  # rank vector in np (no dict necessary)
  rank = sorted_vec.mean(axis=1)
  # construct quantile normalized matrix
  return np.array([

u8sand

export class PriorityCache<V = any> {
  store: {
    [key: string]: {
      value: V,
      start: number,
      cost: number,
      demand: number,
      size: number,
    }
  } = {}

u8sand

#!/bin/python
# pip install click Pillow tqdm
# dezoomify.py will infer the shape while downloading all the tiles and then re-assemble a complete image.
# Usage:
#  python dezoomify.py -v -Z 5 -t tmp -o output.jpg https://baseurl/zoomify/id/TileGroup{0,1,2,3}
#   -Z n: zoom level (if not specified, will identify the best zoom level and download that one)
#   -v: verbosity
#   -t [dir]: temporary directory (can be omitted, but recommended in case it needs to be re-run)
#   -o [file.jpg]: output file
#  [url...]: the urls to download the tiles

u8sand

import pandas as pd

ncbi = pd.read_csv('ftp://ftp.ncbi.nih.gov/gene/DATA/GENE_INFO/Mammalia/Homo_sapiens.gene_info.gz', sep='\t')
# Ensure nulls are treated as such
ncbi = ncbi.applymap(lambda v: float('nan') if type(v) == str and v == '-' else v)
# Break up lists
split_list = lambda v: v.split('|') if type(v) == str else []
ncbi['dbXrefs'] = ncbi['dbXrefs'].apply(split_list)
ncbi['Synonyms'] = ncbi['Synonyms'].apply(split_list)
ncbi['LocusTag'] = ncbi['LocusTag'].apply(split_list)

u8sand

// Place your key bindings in this file to overwrite the defaults
[
  {
    "key": "shift+enter",
    "command": "workbench.action.terminal.runSelectedText",
    "when": "editorTextFocus && terminalIsOpen"
  },
  {
    "key": "shift+enter",
    "command": "markdown-preview-enhanced.runCodeChunk",

u8sand

import urllib.parse as urllib_parse

class ParsedUrl:
  ''' A convenient object for manipulating urls -- works
  like urlparse + parse_qs but mutable

  Example:
    url = ParsedUrl('http://google.com/?q=hello+world&a=this#anchor')
    del url.fragment
    del url.query['a']

u8sand

import uuid
U = uuid.UUID('00000000-0000-0000-0000-000000000000')
def canonical_uuid(obj):
  return str(uuid.uuid5(U, str(obj)))

u8sand

''' Usage:
val = {1: [{'hello': 'world', 2: 'okay'}], 2: []}
typeof(val) == dict[int: list,list[dict[int,str: str]]]
typeof(val, 2) == dict[int: list]

i.e. it will look (as deep as you want) into your object and let you know what it is
'''
import functools

def isIterable(v):

u8sand

''' Motivation: standard python collections.defaultdict doesn't work to well with
deep defaultdicts because it only actually creates the object on __setitem__; this
means the following occurs:
```
d = collections.defaultdict(lambda: collections.defaultdict(lambda: []))
d[1][2].append(3)
3 in d[1][2] == False # you would have expected it to be True
```

The above situation works for the below implementation. The trade-off is potential