link to smoove

count total number of aligned reads inside of lumpy_filter

Only include paired-end sequencing reads in discordants file

Change inline functions to static inline

match behavior of extract_sv_reads and samblaster

this also uses @ernfrid 's method of dropping sequence and qualities
from the split and disc bams. This makes the files about half the size
and the memory (of uncompressed bam1_t) savings is dramatic.

It also fixes multi-threading of CRAM and uses minimal (level 1) gzip
compression.

Changes to get NM in output BAM.

You need to decode the sequence to get at NM and MD must be requested

use some cram decode stuff to reduce run-time of lumpy_filter for cram

Make all lumpy python scripts python2.7 and python3 compatible

exclude reads that are unmapped, not primary, failing-qc, duplicate

support cram in lumpy filter

Add info about low complexity exclude regions.

Add samblaster to bwa mem pipe.

filter.c: check return value of sam_read1.

0.2.13 cram support

a -1 indicates EOF. less than that indicates an error so we exit with
non-zero exit status.

lumpy_smooth: default temp directory to $TMPDIR as found by mktemp

test svtyper. allow specifying -n name of vcf output.

lumpy_smooth: dont run lumpy_filter if .disc and .split bams exist

Par filter

also run lumpy_tests in travis

closes #41

Update README.md