DOKK / manpages / debian 11 / snap-aligner / snap-aligner-paired.1.en
SNAP-ALIGNER_PAIRED(1) User Commands SNAP-ALIGNER_PAIRED(1)

snap-aligner_paired - scalable nucleotide alignment program

Welcome to SNAP version 1.0.0.

Too few parameters Usage: snap-aligner paired <index-dir> <inputFile(s)> [<options>] where <input file(s)> is a list of files to process.

filename output alignments to filename in SAM or BAM format, depending on the file extension or explicit type specifier (see below). Use a dash with an explicit type specifier to write to stdout, so for example -o -sam - would write SAM output to stdout
maximum edit distance allowed per read or pair (default: 27)
number of seeds to use per read
Seed coverage (i.e., readSize/seedSize). Floating point. Exclusive with -n. (default uses -n)
maximum hits to consider per seed (default: 300)
minimum seed matches per location (default: 1)
number of threads (default is one per core)
Don't bind each thread to its processor (note the double dash)
disables cache prefetching in the genome; may be helpful for machines with small caches or lots of cores/cache
sort output file by alignment location
memory to use for sorting in Gbytes. Default is 1 Gbyte/thread.
Specifies the sort intermediate directory. When SNAP is sorting, it aligns the reads in the order in which they come in, and writes the aligned reads in batches to a temporary file. When the aligning is done, it does a merge sort from the temporary file into the final output file. By default, the intermediate file is in the same directory as the output file, but for performance or space reasons, you might want to put it elsewhere. If so, use this option.
explore some hits of overly popular seeds (useful for filtering)
suppress additional processing (sorted BAM output only) i=index, d=duplicate marking
stop on first match within edit distance limit (filtering mode)
filter output (a=aligned only, s=single hit only (MAPQ >= 10), u=unaligned only, l=long enough to align (see -mrl))
an alternate (and fully general) way to specify filter options. Emit only these types s = single hit (MAPQ >= 10), m = multiple hit (MAPQ < 10), x = not long enough to align, u = unaligned, b = filter must apply to both ends of a paired-end read. Combine the letters after -E, so for example -E smu will emit all reads that aren't too short/have too many Ns (because it leaves off l). -E smx is the same as -F a, -E ux is the same as -F u, and so forth. When filtering in paired-end mode (either with -F or -E) unless you specify the b flag a read will be emitted if it's mate pair passes the filter Even if the read itself does not. If you specify b mode, then a read will be emitted only if it and its partner both pass the filter.
ignore IDs that don't match in the paired-end aligner

-Cxx must be followed by two + or - symbols saying whether to clip low-quality

bases from front and back of read respectively; default: back only (-C-+)
-=
use the new style CIGAR strings with = and X rather than M. The opposite of -M
specify the name of a file to contain the run speed
Indicates to use huge pages (this may speed up alignment and slow down index load).
Specifies the extra search depth (the edit distance beyond the best hit that SNAP uses to compute MAPQ). Default 1
Specify the default read group if it is not specified in the input file
Specify the entire read group line for the SAM/BAM output. This must include an ID tag. If it doesn't start with '@RG' SNAP will add that. Specify tabs by \t. Two backslashes will generate a single backslash. backslash followed by anything else is illegal. So, '-R @RG\tID:foo\tDS:my data' would generate reads with default tag foo, and an @RG line that also included the DS:my data field.
Include reads from SAM or BAM files with the secondary (0x100) or supplementary (0x800) flag set; default is to drop them.
Output multiple alignments. Takes as a parameter the maximum extra edit distance relative to the best alignment to allow for secondary alignments
This means that if -om would generate more
than -omax secondary alignments, SNAP will write out only the best -omax of them, where 'best' means 'with the lowest edit distance'. Ties are broken arbitrarily.

-mpc Limit the number of alignments generated by -om to this many per contig (chromosome/FASTA entry);

'mpc' means 'max per contig; default unlimited.
This filter is applied prior to -omax. The primary alignment
is counted.
Preserve the soft clipping for reads coming from SAM or BAM files
Increase expansion factor for BAM and GZ files (default 1.0)

-hdp Use Hadoop-style prefixes (reporter:status:...) on error messages, and emit hadoop-style progress messages

This should be
Default 50
This might speed up index loading in cases
where SNAP is run repatedly on the same index, and the index is larger than half of the memory size of the machine. On some operating systems, loading an index with -map is much slower than without if the index is not in memory. You might consider adding -pre to prefetch the index into system cache when loading with -map when you don't expect the index to be in cache. This is the default

-map- Do not map the index, read it using standard read/write calls.

This is only meaningful with -map, and only helps if the index is not
already in memory and your operating system is slow at reading mapped files (i.e., some versions of Linux, but not Windows). This is the default on Linux.
This is the default on Windows.
Run SNAP at low scheduling priority (Only implemented on Windows)
No Ukkonen: don't reduce edit distance search based on prior candidates. This option is purely for evaluating the performance effect of using Ukkonen's algorithm rather than Smith-Waterman, and specifying it will slow down execution without improving the alignments.
No Ordering: don't order the evalutation of potential alignments so as to select more likely candidates first. This option is purely for evaluating the performance effect of the read evaluation order, and specifying it will slow down execution without improving alignments.
Don't truncate searches based on missed seed hits. This option is purely for evaluating the performance effect of candidate truncation, and specifying it will slow down execution without improving alignments.
Write buffer size in megabytes. Don't specify this unless you've gotten an error message saying to make it bigger. Default 16.
Drop the index after aligning and before sorting. This frees up memory for the sort at the expense of not having the index loaded for your next run.
Kill if too slow. Monitor our progress and kill ourself if we're not moving fast enough. This is intended for use on machines with limited memory, where some alignment tasks will push SNAP into paging, and take disproportinaltely long. This allows the script to move on to the next alignment. Only works when generating output, and not during the sort phase. If you're running out of memory sorting, try using -di.
Profile alignment to give you an idea of how much time is spent aligning and how much waiting for IO

-proAg Profile affine-gap scoring to show how often it forces single-end alignment

Apply the end-of-contig soft clipping before the -om processing rather than after it. A read that's soft clipped because of hanging off one end or the other of a contig does not have a penalty in its NM tag, but it does in SNAP's internal scoring. This flag says to use the NM value for -om processing rather than SNAP's internal score.
Write SNAP's internal score for an alignment into the output. The value following -is specifies the tag to use, and must be a two letter value starting with X, Y or Z. So, -is ZQ will cause SNAP to write ZQ:i:3 on a read with internal score 3. Generally, the internal scores are the same as the NM values, except that they contain penalties for soft clipping reads that hang over the end of contigs (but not for soft clipping that's due to # quality scores or that was present in the input SAM/BAM file and retained due to -pc)
disable affine gap scoring (default: true) Scoring parameters (works only when -G- is not used)
cost for match -gm (default: 1) cost for substitution -gs (default: 4) cost for opening a gap -go (default: 6) cost for extending a gap -ge (default: 1)
Disable ALT awareness. The default is to try to map reads to the primary assembly and only to choose ALT alignments when they're much better, and to compute MAPQ for non-ALT alignments using only non-ALT hits. This flag disables that behavior and results in ALT-oblivious behavior.
Emit ALT alignments. When the aligner is ALT aware (i.e., -A- isn't specified) if it finds an ALT alignment that would have been the primary alignment if -A- had been specified but isn't without -A-, SNAP will emit the read with the supplementary alignment flag set and MAPQ computed across all potential mappings, both primary and ALT
If the best non-ALT alignment is more than this much worse than the best ALT alignment
emit the ALT alignment as the primary result rather than as a supplementary result. (default: 16)

You may process more than one alignment without restarting SNAP, and if possible without reloading the index. In order to do this, list on the command line all of the parameters for the first alignment, followed by a comma (separated by a space from the other parameters) followed by the parameters for the next alignment (including single or paired). You may have as many of these as you please. If two consecutive alignments use the same index, it will not be reloaded. So, for example, you could do 'snap-aligner single hg19-20 foo.fq -o foo.sam , paired hg19-20 end1.fq end2.fq -o paired.sam' and it would not reload the index between the single and paired alignments. When specifying an input or output file, you can simply list the filename, in which case SNAP will infer the type of the file from the file extension (.sam or .bam for example), or you can explicitly specify the file type by preceding the filename with one of the

following type specifiers (which are case sensitive):

-fastq

-compressedFastq

-sam

-bam

-pairedFastq

-pairedInterleavedFastq

-pairedCompressedInterleavedFastq

So, for example, you could specify -bam input.file to make SNAP treat input.file as a BAM file, even though it would ordinarily assume a FASTQ file for input or a SAM file for output when it doesn't recoginize the file extension. In order to use a file name that begins with a '-' and not have SNAP treat it as a switch, you must explicitly specify the type. But really, that's just confusing and you shouldn't do it. Input and output may also be from/to stdin/stdout. To do that, use a - for the input or output file name and give an explicit type specifier. So, for example, snap-aligner single myIndex -fastq - -o -sam - would read FASTQ from stdin and write SAM to stdout.

min and max spacing to allow between paired ends (default: 0 1000). If it can't find an alignment in that range, it will run both reads through the single-end aligner.

-ins Infer inter-read spacing by periodially looking at the observed distances

force spacing to lie between min and max.
max hits for intersecting aligner (default: 4000).

-mcp specifies the maximum candidate pool size (An internal data structure.

Only increase this if you get an error message saying to do so. If you're running out of memory, you may want to reduce it. Default: 1000000)

-F b additional option to -F to require both mates to satisfy filter (default is just one)

If you specify -F b together with one of the other -F options, -F b MUST be second
Keep unpaired-looking reads in SAM/BAM input. Ordinarily, if a read doesn't specify mate information (RNEXT field is * and/or PNEXT is 0) then the code that matches reads will immdeiately discard it. Specifying this flag may cause large memory usage for some input files, but may be necessary for some strangely formatted input files. You'll also need to specify this flag for SAM/BAM files that were aligned by a single-end aligner.
max seeds when falling back to the single-end mode when doing paired-end. Default: 25
min edit distance for a read aligned as non-ALT by the paired-end aligner to be reconsidered for a better alignment by the single-end aligner. Default: 3
min total edit distance by which a read pair aligned as ALT needs to be better than non-ALT alignments to skip single-end realignment. Default: 3
min affine gap score improvement needed for single-end alignments to be considered over paired-end alignments. Default: 15
January 2021 snap-aligner paired 1.0.0