compare_ai_and_ncls.py

# wget http://big.databio.org/example_data/AIList/AIListTestData.tgz

nrows = 1.5e6

from ncls import NCLS
from ailist import AIList

import numpy as np
import pandas as pd

from time import time

query_file = "chainRn4.bed"

q = pd.read_csv(query_file, usecols=[1, 2], header=None, dtype=np.int64, nrows=nrows, sep="\t")

def time_ncls(s, q):

    start = time()
    n = NCLS(s[1].values, s[2].values, s.index.values)
    end = time()

    print("Building NCLS took", end - start)

    n.all_overlaps_both(q[1].values, q[2].values, q.index.values)
    end2 = time()
    print("Querying NCLS took", end2 - end)


def time_ai(s, q):

    start = time()
    i = AIList()
    i.from_array(s[1].values, s[2].values, s.index.values, np.array(s.index.values, dtype=np.double))
    i.construct()
    end = time()
    print("Building ai took", end - start)

    i.intersect_from_array(q[1].values, q[2].values, q.index.values)
    end2 = time()
    print("Querying ai took", end2 - end)


for subject_file in "chainOrnAna1.bed chainVicPac2.bed chainXenTro3Link.bed chainMonDom5Link.bed".split():

    print("------" * 5)

    s = pd.read_csv(subject_file, usecols=[1, 2], header=None, dtype=np.int64, nrows=nrows, sep="\t")
    info_string = f"Subject file: {subject_file} ({len(s)/1e6}), Query file: {query_file} ({len(q)/1e6})"
    print(info_string)

    # print(s.head())
    # print(s.dtypes)

    time_ncls(s, q)
    #time_ai_endre(s, q)
    time_ai(s, q)

 # ------------------------------
# Subject file: chainOrnAna1.bed (1.5), Query file: chainRn4.bed (1.5)
# Building NCLS took 0.37735605239868164
# Querying NCLS took 92.85288500785828
# Building ai took 0.3491032123565674
# Querying ai took 75.86743712425232
# ------------------------------
# Subject file: chainVicPac2.bed (1.5), Query file: chainRn4.bed (1.5)
# Building NCLS took 0.39214491844177246
# Querying NCLS took 92.57744908332825
# Building ai took 0.3954188823699951
# Querying ai took 73.98339509963989
# ------------------------------
# Subject file: chainXenTro3Link.bed (1.5), Query file: chainRn4.bed (1.5)
# Building NCLS took 0.3832888603210449
# Querying NCLS took 33.25918006896973
# Building ai took 0.25615978240966797
# Querying ai took 27.772515296936035
# ------------------------------
# Subject file: chainMonDom5Link.bed (1.5), Query file: chainRn4.bed (1.5)
# Building NCLS took 0.26062488555908203
# Querying NCLS took 10.006813049316406
# Building ai took 0.20964312553405762
# Querying ai took 4.56371283531189