Welcome to the MELT homepage.
The Mobile Element Locator Tool (MELT) is a software package, written in Java, that discovers, annotates, and genotypes non-reference Mobile Element Insertions (MEIs) in Illumina DNA paired-end whole genome sequencing (WGS) data. MELT was first conceived as a tool to identify non-reference MEIs in large genome sequencing projects, specifically as part of the 1000 Genomes Project, and has been further optimized to run on a wide range of data types. MELT is optimized for performing discovery in a large number of individual sequencing samples using the Sun Grid Engine (SGE). MELT also has two additional workflows: analysis without SGE (for adaptability to other parallel computing platforms) and single genome analysis. MELT is highly scalable for many different types of data, and the different workflows are outlined and detailed in this documentation.
- First new public release in over a year - the 2.2.0 release incorporates all changes from 2.1.5 to now.
- MELT no longer writes a chromosome mask file in ./MELTWorkingDir/tmp/ directory. It is now stored in memory.
- Internal changes to code for Alignment to MEI references. Functionality is same as previous versions.
- Fixed a bug related to where, on rare occasions, MELT would fail during GroupAnalysis due to not being able to find discordant pairs at an identified locus.
- Added a new error message that actually makes sense for when MELT cannot find an index for a BAM/CRAM file.
- Internal changes to how MELT handles the transposon reference mask.
- Small optimization for CRAM use.
- Added DP and AD tags to the MELT VCF file. These report the number of reads that MELT assessed and number of these reads that supported the insert site, respectively.
- Set the -c options to “not required”. MELT will now calculate coverage on it's own from the first chromosome in the .fai index starting at the value of -d (so default position 1,000,000). This calculated value can be overriden by either setting -c or putting a second column in the bamlist file as before.
- Fixed a bug where the -exome flag was not working with MELT Single (thanks to verne91 for reporting).
- Implemented htsjdk version (2.16.0)
- MELT now supports CRAM.
- Please note, to use CRAM format files you will need a working 'samtools' binary either in your path or supplied via the -samtools option.
- MELT has been tested with the latest changes to ensure results are identical for BAM files between MELT versions. This has not been completely verified for CRAM format files.
- Various under the hood movement in the MELT source code to make it easier to maintain.
- Changed slightly how gene reporting works within MELT. Will now report ALL overlaps with file provided by -n seperated by |.
- Several small bug fixes.
- Fixed bug in MELT-Deletion that caused issues when the reference genome was being masked (Thanks M. Okhovat for bug report).
- Added the '-sr' flag for filtering reads without SR support during the GroupAnalysis step.
- Added SR flag to the MELT MEI VCF output which displays the total number of SRs per site.
- Fixed a bug in the genotyping step that caused organisms with small chromosomes/contigs to fail at the genotyping step (Thanks to M. Volpe for reporting).
- Implemented new htsjdk version (2.11.0).
- To keep MELT current with relevant libraries in preperation for additional available datasets.
- Bug Fixes:
- Removed old code supporting multi-threaded discovery.
- Fixed an issue where MELT would fail if the BAM file header from the source BAM was too large (Thanks M. Volpe for reporting).
- Fixed an issue where sites would fail on estimating the breakpoint due to lack of high quality reads (Thanks to R. Collins for reporting).
- Fixed an issue where MEIs being discovered at coordinate “0” on any given contig where resulting in a genotype failure (Thanks to M. Volpe for reporting).
- First post-publication update to MELT.
- Fixed a bug in MELT-DEL which would cause MELT to freeze when dealing with sites immediately adjacent to ’N’ sequence in reference genomes (Thanks J. Korstian for original bug report).
- Fixed a bug in MELT-DEL where ‘./.’ genotypes would be reported if no ‘MD’ tag was present to determine mismatches.
- Fixed a bug where excessive supplementary alignments were causing MELT to have excessive runtimes during the IndivAnalysis step.
- Added a new flag when running MELT with exome data: -exome
- This flag changes the minimum amount of evidence necessary to place a site in the putative hits list during IndivAnalysis
- Changed how MELT handles tmp directories to avoid confusing errors when running a large number of files at the same time through MELT-Split.
- Changed number of records merged into the final VCF file from 5,000 to 1,000 to better accommodate large datasets.
- Added the “ac0” FILTER for sites that did not genotype in the final VCF merge.
- MELT no longer filters AF = 0 sites when making the final VCF.
- This can be disabled with the 'ac' flag.
- Improved scalability of the MELT genotyper when assessing excessively high coverage genomes (e.g. > 60x) to remove false positive sites.
- Added a filter (lc) in the final MELT vcf for MEIs abutting low complexity regions.
- This does not filter sites automatically, but flags them as lower confidence sites.
- Functions by scanning ±25bp from the breakpoint and looking for runs of 15 or more bases of A,T,C,G or any combination of the two (i.e. an AT micro satellite).
- WARNING: This filter has not been testing using Whole Genome data yet (only exome) and may lead to lower sensitivity in relevant datasets.
- Implemented a new Genotyping optimization for short elements (i.e. SINEs) that improves het -> hom genotypes.
- Internal changes in how MELT handles options.
- MELT now has a single set of options that are used by all MELT pipelines.
- Excludes CAlu, LINEU, and BuildTransposonZIP.
- Internal changes on how MELT performs analysis in MELT-SGE and MELT-Single modes.
- Fixed a bug that caused the wrong max-coverage filter to be used in MELT-Split mode.
- Fixed a bug that was incorrectly handling the ‘-a’ flag (BWA-aln was used for alignment).
- Changed the transduction module to three parts.
- Use ‘Source’ to generate a list of source L1s for transduction analysis.
- Can now preprocess bam files for transductions. This should hopefully increase the speed of transduction finding.
- Now use TransductionMerge of a list of bam files to finalize VCF annotation.
- Changed the output in the MELT VCF INFO field for transductions.
- MESOURCE now has the format <SOURCE CHR>,<SOURCE POS>,<SOURCE TYPE [REF || nonREF]>,<GENOTYPE CONCORDANCE>,<TOTAL SUPPORTING READS>,<TRANSDUCTION SIZE>
- Additional INFO field METRANS has the info <CHR>,<TRANSDUCTION SOURCE START>,<TRANSDUCTION SOURCE STOP>,<TRANSDUCTION STRAND>.
- See MELT manual for more details.
- Fixed a bug where MELT would report a 5’ Inversion event rather than reporting no measurement (-1).
- Slight efficiency improvements in step 1 of the MELT-Split pipeline.
- Added better description/explanation for the error reported in GroupAnalysis when MELT doesn’t find any hits in the IndivAnalysis step.
- Added check to the MakeVCF step to ensure all genotyping finished properly.
- Fixed error in MELT transduction finder due to unrecognized bam names.
- Fixed an error in MELT genotype where MEI masking was not working properly.
- Added a format check for the genome annotation file (provided with -n).
- Updated some language in the help dialog provided by --help/-h.
- Several small bug fixes.
- Fixed an error in the manual that pointed to the wrong human reference file to download in the 'MELT Quick-Start Guide' section
- Initial Release
- Several improvements beyond MELT v1.0 used for the 2015 Nature publication, see our paper for more details.