Daniel Nicorici ndaniel

This is a small experiment on the alignment of ~50bp INDELs. The query sequences are shown in 0.01.fq below, where seq_ori is a 204bp sequence extracted from the human reference genome, seq_del54 contains a 54bp deletion in the middle, seq_del84 contains a 84bp deletion in a 120bp read, and seq_ins40 contains a 40bp insertion in a 140bp read. These four short sequences were mapped to the human reference genome with Bowtie2, BWA-MEM, LAST, Novoalign, SNAP and Stampy with default settings. Non-default scoring functions were also tested for Bowtie2 (--rdg 5,1 --rfg 5,1), BWA-MEM (-A2 -E1) and LAST (-r2 -q4). The output by various mappers/settings can be found in this gist. The following table gives my summary:

Mapper	Setting	-84bp	-54bp	+40bp
BBMAP	default	Yes	Yes	Yes
Bowtie2	default	No	No	No
Bowtie2	--rdg 5,1 --rfg 5,1	as insertion	as insertion	Yes
BWA-MEM	default	as split	Yes	Yes
BWA-MEM	-A2 -E1	Yes	Yes	Yes
LAST	default	as split	as split

	"""making a dataframe"""
	df = pd.DataFrame([[1, 2], [3, 4]], columns=list('AB'))

	"""quick way to create an interesting data frame to try things out"""
	df = pd.DataFrame(np.random.randn(5, 4), columns=['a', 'b', 'c', 'd'])

	"""convert a dictionary into a DataFrame"""
	"""make the keys into columns"""
	df = pd.DataFrame(dic, index=[0])

	# Try this with:
	# - https://github.com/jefworks/genesets
	# - https://github.com/slowkow/tftargets


	#' Compute a hypergeometric p-value for your gene set of interest relative to
	#' a universe of genes that you have defined.
	#'
	#' @param ids A vector with genes of interest.
	#' @param universe A vector with all genes, including the genes of interest.