Fast way of representing mapped read positions and comparing two sets of mapped reads using numpy. This package can make nice report pretty fast, comparing mapping accuracies of different sets of alignments.
Clone repo and enter directory, then run:
pip3 install ... or just pip install it (Python3):
pip3 install numpy_alignmentsImportant: Numpy alignments always needs to know the number of reads when it is run. This package is specifically made to work with simulated reads, and requires all reads to have names from 0 to the number of reads (such as reads simulated by Graph Read Simulator). This makes it able to efficiently represent the reads in numppy arrays of the given size. The following examples assume there are 100 reads.
Map reads with BWA-MEM and pipe directly to numpy alignments to avoid storing large BAM-files on disk:
bwa mem ref.fa reads.fa | numpy_alignments store sam bwa 100Save truth positions:
cat positions.tsv | numpy_alignments store truth truth 265154Compare bwa to truth:
numpy_alignments get_correct_rates truth bwaCreate html report:
numpy_alignments make_report -f my-report-name --names="bwa" truth bwa purpleYou can specify more names, alignments sets and colors by separating them with commas.
from numpy_alignments.comparer import Comparer
from numpy_alignments import NumpyAlignments
bwa = NumpyAlignments.from_file("bwa.npz")
truth = NumpyAlignments.from_file("truth.npz")
comparer = Comparer(truth, {"bwa": bwa})
rate = comparer.get_correct_rates()
print(rate)