seqan / lambda

LAMBDA – the Local Aligner for Massive Biological DatA

SeqAn - The Library for Sequence Analysis

NOTE SeqAn3 is out and hosted in a different repository
We recommend using SeqAn3 for new applications.

What Is SeqAn?

SeqAn is an open source C++ library of efficient algorithms and data structures for the analysis of sequences with the focus on biological data. Our library applies a unique generic design that guarantees high performance, generality, extensibility, and integration with other libraries. SeqAn is easy to use and simplifies the development of new software tools with a minimal loss of performance.

License

The SeqAn library itself, the tests and demos are licensed under the very permissive 3-clause BSD License. The licenses for the applications themselves can be found in the LICENSE files.

Prerequisites

Older compiler versions might work but are neither supported nor tested.

Linux, macOS, FreeBSD

• GCC ≥ 11
• Clang/LLVM ≥ 15
• Intel oneAPI C++ Compiler 2024.0.2 (IntelLLVM)

Windows

• Visual C++ ≥ 17.0 / Visual Studio ≥ 2022

Architecture support

• Intel/AMD platforms, including optimisations for modern instruction sets (POPCNT, SSE4, AVX2, AVX512)
• All Debian release architectures supported, including most ARM and all PowerPC platforms.

Build system

• To build tests, demos, and official SeqAn applications you also need CMake ≥ 3.12.

Some official applications might have additional requirements or only work on a subset of platforms.

Documentation Resources

• Getting Started
• Manual
• Tutorial
• How-Tos
• API Documentation (stable)

Contact

• Mailing List
• GitHub Project (issues, source code)

lambda's Issues

update documentation

• update README.md and -> README.rst
• CHANGELOG.md -> CHANGELOG.rst -> wiki
• LICENSE -> LICENSE.rst
• update documentation in old lambda branch to point here
• add wiki-pages
• add homepage
• make files, manual and homepage link to the wiki

speed-up FM-Index creation

by using ModifiedStrings internally instead of copying.

better diagnosis of index compatibility

• general improvement

and another bug :

• BLASTN indexes are now always classified as too old...
• add return value to indicate old index

reduce compile time and binary size

• use dynamic score matrix
• strip binaries on gcc
• scoring optimizations for linear gap costs hidden behind extra macro and deactivated by default
• removed SimpleScore for aminoacid searches

Discard early matches of XXXXXX

Never search with seeds that contain more then a certain number of Xs.

Port to SeqAn 2.0 and new Blast IO

• make code build
• make code run; via ef6d62e
• upstream necessary changes to SeqAn
• test and benchmark

using seg files is broken

crash when putative duplicates heuristic is off

Add app-tests

run clang static code analyzer...

... and possibly fix stuff. Maybe it add it to travis, as well?

re-extend alignment if it previously went out-of-band

develop a heuristic to re-extend an alignment, if the previous extension likely resulted in a failure, because of the band limitation.

hide most options by default

the huge list of options is too confusing

Support for custom Columns in Blast Tabular Out

crash in tblastn mode

Make database sizetype automatically detectable

also relates to #7

parallel builds on GCC

automatically use -flto=n on GCC to parallelize build even with one translation unit [ in 48d3ead ]

support for openbsd

only do traceback if score is good enough

• for SIMD extensions via cd0bd41 f.
• for serial extensions
• for xdrop (doesnt make sense)

-nm not connected to internal variable

fixed by h-2/seqan@f639f2a

Improve indexing algorithms

• add radixsort sort from Sascha Meiers, ad2ee38
• benchmark on larger data
• clean up files and throw out other sorting algorithms
• import BPR and BCR, see seqan/seqan#683 and seqan/seqan#685
  Possibly snatize the SeqAn interface on the way and add easy option to select algorithm at run-time.

Develop new format for database index

• for uniprot_sprot.fasta create uniprot_sprot.fasta.lambda directory
• all currently created files will be moved to this directory
• possibly create an options file or individual files indicating certain parameter choices when creating index

always warn when not running in debug mode

to prevent issues like during 0.9.0 release

improve progress reporting while indexing

i.e. seperate progress for SA and FM

new murphy10 might have reduced sensitivity

... because * is clustered with FWY instead of A...?

support Clang as compiler

add supprt for clang

Implement partitioning of files

Possibly implement proper partitioning of query and/or subject files, with lazy loading.

use mmapped IO in other places

• Evaluate the benefits of using mmapped IO for other sequences and even SA
• possibly develope automatic chosing of different sequence types depending on available RAM
• also relates to #7

variable length seeding

moving target in 1.9.x cycle

deprecate masking

detect if the index is compatible

... and exit otherwise.

the indexer has becoming notably slower

at least on the linux-servers.

Use mmaped IO for entire database

* make ot optional and off by default for now, because it breaks compatibility to old indeces

• entire database read through mmapped IO, f4ef9ad
• make default and build flag to turn this off, 3b7a6fa
• benchmark influence on performance

add more tests

• protein but not alphabet reduction
• seg
• …

Add support for bam and sam

make accessing read-only indexes work

see seqan/seqan#507

error in help-page

Achja, noch ein Hinweis: Der help-String von Lambda im Feature-Branch
ist nicht ganz konsistent: du beschreibst die Parameter für den
Fast-Mode mit "-ar 0 -sl 7 -sd 0", es müsste bei der aktuellen
Programmversion aber "-ar murphy10" sein.

segfault on TBlastX mode

https://github.com/h-2/seqan/blob/feature/lambda/extras/apps/lambda/holders.hpp#L400

will access element behind the last when subject sequences are translated, because n entries are created for masking, but after translation 6*n elements exist. TODO escape this entire for-loop if no masking is used.

Perform some sanity checks on input

Make sure sequences are in right alphabets and catch io-exceptions.

crash with blastn in combination with suffix array index

add static libgcc on mac

although static builds are not possible, it seems to be possible to add -static-libgcc -static-libstdc++ which already improves the portability of the binaries.

make sure gzipped and bzipped IO works

add option to truncate subject ids on index creation

... to save space.

The full IDs are only needed for .m0 output.

results not the same with old version

1-5% less results.

apperently not related to alphabet.

continuous integration

travis ci

add option for soft-clipping to SAM/BAM

iterative mode

double check default build type

someone reported that it was Debug and not Release.

Develop non-binary wavelet tree FM index

• implement in SeqAn
• integrate in Lambda

support icc as compiler

add os x to travis