cram-speed.nim

import hts
import strutils
import times
import os


const everything = int(8191)

proc timeit(path:string, fasta:string, threads:int, except_fields:varargs[SamField]) =
  var
    cram: Bam
  open(cram, path, fai=fasta, threads=threads)
  var exclude:int
  var names = newSeq[string]()
  for e in except_fields:
      exclude = exclude or e.int
      var name = $e
      names.add(name[4..name.high])
  if names.len == 0: names = @["nothing"]
  if exclude != 0:
    if cram.set_option(FormatOption.CRAM_OPT_REQUIRED_FIELDS, everything - exclude) != 0:
      quit "couldn't set option:"

  var t = epochTime()
  var n = 0

  for aln in cram:
      n += 1
  echo join(names, "|"), "\t", threads, "\t", epochTime() - t, "\t", n

var
  path = commandLineParams()[0]
  fasta = commandLineParams()[1]


echo "exclude\tthreads\ttime\treads"
for threads in @[0, 1, 2, 4]:
  timeit(path, fasta, threads)
  timeit(path, fasta, threads, SAM_SEQ)
  timeit(path, fasta, threads, SAM_QUAL)
  timeit(path, fasta, threads, SAM_QNAME, SAM_RNAME)
  timeit(path, fasta, threads, SAM_RNAME, SAM_AUX, SAM_RGAUX, SAM_RNEXT, SAM_PNEXT, SAM_TLEN)
  timeit(path, fasta, threads, SAM_QNAME, SAM_SEQ, SAM_QUAL)
  timeit(path, fasta, threads, SAM_QNAME, SAM_SEQ, SAM_QUAL, SAM_AUX, SAM_RGAUX)
  timeit(path, fasta, threads, SAM_QNAME, SAM_SEQ, SAM_QUAL, SAM_AUX, SAM_RGAUX, SAM_RNEXT, SAM_PNEXT, SAM_TLEN)

plot-cram-speed.py

from matplotlib import pyplot as plt
import seaborn as sns

import pandas as pd
import sys

colors = sns.color_palette()

df = pd.read_table(sys.argv[1])


df.exclude = [x.replace("|", "\n") for x in df.exclude]

ax = sns.barplot(x="exclude", y="time", hue="threads", data=df)
ax.set_xlabel("Exclude Fields")
ax.set_ylabel("Time (Seconds)")

for t, sd in df.groupby('threads'):
    baseline = sd.time[sd.exclude == "nothing"]
    #ax.axhline(y=float(baseline)/3.0, color=colors[t])

#ax.text(0.7, 0.8, "horizontal lines show height of 3X speed improvement over baseline")
plt.tight_layout()
plt.show()