Tools for processing DALIGNER .las files
The lastools source code is hosted on github:
git@github.com:gt1/lastools.git
lastools needs libmaus2 [https://github.com/gt1/libmaus2] . When libmaus2
is installed in
- autoreconf -i -f
- ./configure --with-libmaus2=${LIBMAUSPREFIX} \
--prefix=${HOME}/lastools
- make install
The command line parameters of all programs can be obtained by calling the respective program with the --help switch as only parameter. The contained programs are:
- bamtolas: Convert an input BAM file to the LAS format. This only converts the data which can be stored in LAS files, i.e. no auxiliary tags etc.
- fasta2fastq: Convert an input FastA file to FastQ by adding dummy quality values ('H').
- lasindex: Produce an index for one or more LAS files.
- lasroleswap: Swap the roles of the A and B read in an LAS file.
- lassort: Sort a set of LAS files and merge into a single output file.
- lastobam: Convert a sorted input LAS file to the BAM format. This does currently not add information about unmapped reads.
- laschainsort: Chain aware sorting of LAS files. Use this for sorting LAS files produced by damapper.
- reformatfasta: Reformats a FastA file so it follows the naming and formatting (limited column width) of the FastA expected by fasta2DB in the DAZZ_DB suite.
- call.damapper: A wrapper program for damapper and lastobam. It takes a reference and a read FastA file and produces a BAM file containing the reads as mapped by damapper.
- lassubsample: Subsample an LAS file to retain only a given fraction of reads.
- viewmasks: View repeat and tandem masks produced by damasker package
The lastobam program requires the A reads in LAS files to refer to the reference sequence database and the B reads to the read sequence database. This is not the default output format of damapper, so make sure to call damapper (or HPC.damapper) using the the switches -C -N. lastobam expects to see the alignments in increasing order of the B read id. Use laschainsort with switch -sba to obtain this order. A sample pipeline is thus:
# k-mer size used by damapper
k=20
# create dazzler DAM file for reference
fasta2DAM ref.dam ref.fasta
# call DBsplit (block size 250MB, cut off at the kmer size)
DBsplit -s250 -x${k} ref.dam
# create dazzler DAM file for the reads
fasta2DAM reads.dam reads.fasta
# call DBsplit on the reads with block size 250MB, cut nothing off
# (-x0 is important because lastobam cannot handle trimmed read databases)
DBsplit -s250 -x0 reads.dam
# run damapper
HPC.damapper -C -N -k${k} <other switches> ref.dam reads.dam | grep "^damapper" | ${SHELL}
# sort
laschainsort -sba sorted.las ref.reads.las
# convert the alignments to bam format
lastobam -snone ref.dam ref.fasta reads.dam reads.fasta sorted.las >out.bam
call.damapper is a simplified interface to damapper and lastobam. An example call is
call.damapper ref.fasta <reads.fasta >reads.bam
call.damapper requires the programs fasta2DAM, DBsplit (both from https://github.com/thegenemyers/DAZZ_DB), HPC.damapper, damapper (both from https://github.com/thegenemyers/DAMAPPER), lascat and lastobam (both in this repository) either accessible via PATH or in the same directory as call.damapper.
call.damapper passes the parameters k,t,M,e,s,n,B,T,b,v and p through to damapper. Please see https://github.com/thegenemyers/DAMAPPER and https://github.com/thegenemyers/DALIGNER for their meaning. Some of the more frequently relevant ones are
- -T: number of threads (by default -T4)
- -M: memory limit (by default the machine's memory size)
- -k: k-mer size (by default -k20)
Additionally call.damapper has the options
- --refblocksize: reference block size used for DBsplit. The default is --refblocksize256, which sets the block size to 256MB.
- --readblocksize: read block size used for DBsplit. The default is --readblocksize256, which sets the block size to 256MB.
- -I: directory used for storing the reference index. By default the index is stored in the directory containing the reference FastA file.
- -W: working directory. By default the current directory is used for storing temporary files.
Whitespace is not allowed between key and value when providing parameters (e.g. -Idir is valid, -I dir is not).
Please see https://github.com/gt1/damapper_bwt for special aux fields produced in the BAM output by lastobam/call.damapper/damapper_bwt.
daligner (https://github.com/thegenemyers/DALIGNER) usually does a very good job at computing all significant pairwise local alignments in a long read set. However it sometimes does miss some true alignments in tandem repeat regions. The purpose of the following pipeline is to compute (most) of these missing alignments. This requires the program datander, which is contained in the DAMASKER suite (see https://github.com/thegenemyers/DAMASKER). The expected input is a dazzler database (reads.db) and a dazzler alignment file (reads.las) produced by daligner for this database (reads.db).
# call datander, this produces TAN.reads.las
datander reads.db
# call lascomputetandem to compute tandem regions on reads
lascomputetandem TAN.reads.las >TAN.reads.las.intv
# call lasextracttandemrecompute to compute read pair ids
# for which we want compute alignments
# this is done by scanning reads.las and checking for
# reads which overlap with (suspected) tandem repeat
# regions
lasextracttandemrecompute TAN.reads.las.intv reads.las >TAN.reads.las.recomp
# compute alignments in tandem repeat regions
# This will be slower than daligner, as it uses a more
# exhaustive approach.
tandemaligner reads_tandem.las reads.db TAN.reads.las.recomp