Development manual for vsearch 2.32.0, built from the dev branch: it describes changes that are not released yet. The manual for the current release is published separately.

NAME

vsearch --chimeras_denovo — detect chimeras de novo in long exact sequences

SYNOPSIS

vsearch --chimeras_denovo inputfile (--chimeras | --nonchimeras | --alnout | --tabbedout) outputfile [options]

DESCRIPTION

The vsearch command --chimeras_denovo detects chimeras de novo (i.e. without external references) in long exact sequences (inputfile, in fasta or fastq format). It uses a modified UCHIME algorithm that can automatically adapt to a wide range of sequence lengths.

Abundance annotations (pattern [>@;]size=integer[;]) present in sequence headers are taken into account by default. This means that option --sizein is always implied, and does not need to be specified.

Input sequences are automatically sorted by decreasing abundance before detection, and each sequence is compared only against the previously processed, more abundant sequences that were classified non-chimeric. Sequences are sorted into chimeras and non-chimeras, and can be written to fasta files (see output options --chimeras, --nonchimeras). Additional information on each chimera can be collected with the output options --tabbedout and --alnout. The latter outputs for each chimera (i.e. ‘Query’) a multi-way alignment and a model showing the most likely parent sequence of each section of the chimera. Here is an example of a short chimeric sequence (Q), with two parents (A and B):

------------------------------------------------------------------------
Query   (   20 nt) Q
ParentA (   20 nt) A
ParentB (   20 nt) B

Q     1 GTAGGCCGTGCTGAGCCGTA 20
A     1 GTAGGCCGTGgTagGCCGTg 20
B     1 cTgaGCCGTaCTGAGCCGTA 20
Diffs   A AA     AB BB     B
Model   AAAAAAAAAABBBBBBBBBB

Ids.  QA 80.00%, QB 80.00%, QC 0.00%, QT 80.00%, QModel 100.00%, Div. +25.00%

Lowercase positions indicate a mismatch between the parent and the query. The line Diffs indicates the positions that favour a particular parent when modelling the chimera. The line Ids gives global similarity percentages with the different parents (QA, QB, and QC), the closest parent (QT), the model (QModel, always 100.00%), and the divergence of the model with the closest parent (Div). If there are only two parents (A and B), QC is set to 0.00%. If there are more than three parents, only QA, QB and QC are reported.

OPTIONS

mandatory options

--chimeras_denovo inputfile: Detect chimeras de novo in inputfile (fasta or fastq format) using a modified UCHIME algorithm that adapts to a wide range of sequence lengths.

At least one of the following output options must be specified:

--alnout filename

Write multi-way chimera alignments to filename in a human-readable format (see the example in the DESCRIPTION). Use --alignwidth to set the alignment width (default 60 nucleotides).

--chimeras filename

Write chimeric sequences to filename, in fasta format.

--nonchimeras filename

Write non-chimeric sequences to filename, in fasta format.

--tabbedout filename

Write chimera detection results to filename as an eighteen-column tab-separated file, with one row per chimera. Columns are:

score: dummy value, always set to 99.9999
query label
parent A label
parent B label
parent C label (“*” if there are only two parents)
QModel: maximum global similarity percentage (always 100.0%)
QA: global similarity percentage with parent A
QB: global similarity percentage with parent B
QC: global similarity percentage with parent C (0.00 if only two parents)
QT: highest similarity percentage with any parent
left yes: ignored, always set to zero
left no: ignored, always set to zero
left abstain: ignored, always set to zero
right yes: ignored, always set to zero
right no: ignored, always set to zero
right abstain: ignored, always set to zero
dummy value, always set to 0.00
chimeric status, always set to Y (only chimeras are reported)

core options

--abskew real: Set the minimum abundance skew ratio between a chimera and its potential parent. The assumption is that chimeras appear later in the PCR amplification process and are therefore less abundant than their parents. Any positive value equal or greater than 1.0 can be used. Default is 1.0, which means that the parents should be at least as abundant than the chimera.
--chimeras_diff_pct real: Set the maximal mismatch percentage allowed in each chimeric region (excluding insertion and deletions). Accepted values range from 0.0 to 50.0%. Default is 0.0 (no mismatch allowed).
--chimeras_length_min positive non-null integer: Set the minimum length of each chimeric region. Default is 10.
--chimeras_parents_max positive non-null integer: Set the maximum number of parent sequences. Accepted values range from 2 to 20. Default is 3.
--chimeras_parts integer from 2 to 100: Set the number of parts to divide sequences into. Accepted values range from 2 to 100 (inclusive); values outside this range are rejected. The default is one part per 100 nucleotides (rounded up), bounded to the 2 to 100 range.
--sizein: Use the abundance annotations present in sequence headers when reading fasta or fastq file. Search for the pattern [>@;]size=integer[;]. Entries without abundance annotations are silently assumed to be of size=1.; Always implied.

secondary options

--alignwidth positive integer

Set maximal width of three-way alignments when writing alignments with --alnout. Default width is 60 nucleotides. Set to zero (0) to suppress folding.

--fasta_width positive integer

Set the maximal width of sequences when writing fasta files. Longer sequences are folded and written on several lines. Default width is 80 nucleotides. Set to zero (0) to suppress folding.

--hardmask

Replace masked nucleotides with Ns, rather than lowercasing them.

This is also the only masking option that reaches the alignment. Soft and dust masking (see --qmask and --dbmask) only keep masked words out of the k-mer pre-filter that selects candidates; the masked region is still aligned and scored like any other. An N, on the other hand, scores zero and counts as a matching column, so hard masking makes a masked region match whatever it is aligned against, and the reported identity can only go up. Add --n_mismatch to count those columns as mismatches instead.

--label_suffix string

Add the suffix string to sequence headers when writing fasta or fastq files. For example, with --label_suffix ";status=healthy", sequence header ‘>seq1’ becomes ‘>seq1;status=healthy’.

--lengthout

Add a sequence length annotation (;length=integer) to each sequence header when writing fasta or fastq files.

--log filename

Write messages to filename. Messages include program version, start and finish times, elapsed time, amount of memory available, maximum amount of memory consumed, number of cores and command line options, and if need be, command-specific informational messages, warnings, and errors. Messages are written in addition to the standard error, not instead of it; use --quiet to silence the standard error copy (see the --quiet entry for what it spares).

--maxseqlength positive integer

Discard sequences longer than positive integer (50,000 nucleotides by default). The value must not exceed 2,147,481,646 (INT_MAX minus 2,001). When that value is smaller than the effective --minseqlength value (whose default is command-specific), no sequence can pass the length filter: vsearch issues a warning.

--minseqlength non-negative integer

Discard sequences shorter than non-negative integer (1 nucleotide by default). A value of 0 retains empty sequences. When that value is larger than the --maxseqlength value, no sequence can pass the length filter: vsearch issues a warning.

--no_progress

Suppress the gradually increasing progress indicator normally written to the standard error stderr(3).

--notrunclabels

Retain whole sequence headers in output files. By default, vsearch truncates sequence headers at first space or tabulation. This option suppresses truncation.

--qmask none|dust|soft

Mask regions in query sequences using the dust method or the soft method, or none to suppress masking. Values are case-insensitive, so DUST, Dust, and dust are all accepted. See vsearch-fastx_mask(1) for more details. Warning, when using soft masking, search commands become case sensitive: masking excludes masked regions from the k-mer pre-filter that selects candidate targets (the pairwise alignment itself always ignores case). A query with no unmasked stretch of at least the word length samples no k-mers and is therefore compared against every database sequence. The default is to mask using dust.

--quiet

Suppress messages to the standard output stdout(3) and standard error stderr(3), except for warnings and error messages.

--relabel string

Replace sequence headers with the prefix string and a ticker (1, 2, 3, etc.). For example, with --relabel "cluster:", the first sequence header becomes ‘>cluster:1’, the second sequence header becomes ‘>cluster:2’, and so on. To retain annotations, use their corresponding options (--lengthout, --eeout, and --sizeout). Use --relabel_keep to also retain old sequence identifiers.

--relabel_keep

Retain old sequence identifiers by including them at the end of the new headers, after a space.

--relabel_md5

Replace each sequence header with the MD5 digest derived from the sequence itself. The sequence is converted to upper case, and each ‘U’ is replaced with a ‘T’ before computation of the digest. The MD5 digest is a 128-bit value (16 bytes), represented using a string of 32 ASCII characters. Each pair of characters encodes an hexadecimal value, ranging from x00 to xff. See md5(3) for more details, and --relabel_sha1 for an alternative hashing algorithm. To retain annotations, use their corresponding options (--lengthout, --eeout, and --sizeout). Use --relabel_keep to also retain old sequence identifiers.

--relabel_self

Replace each sequence header with the sequence itself. To retain annotations, use their corresponding options (--lengthout, --eeout, and --sizeout). Use --relabel_keep to also retain old sequence identifiers.

--relabel_sha1

Replace each sequence header with the SHA1 digest derived from the sequence itself. The sequence is converted to upper case, and each ‘U’ is replaced with a ‘T’ before computation of the digest. The SHA1 digest is a 160-bit value (20 bytes), represented using a string of 40 ASCII characters. Each pair of characters encodes an hexadecimal value, ranging from x00 to xff. See sha1(3) for more details, and --relabel_md5 for an alternative hashing algorithm. To retain annotations, use their corresponding options (--lengthout, --eeout, and --sizeout). Use --relabel_keep to also retain old sequence identifiers.

--sample string

Add the given sample identifier string to sequence headers when writing fasta or fastq files. For instance, if string is ‘ABC’, the text ;sample=ABC will be added to the headers. string is silently truncated at the first ‘;’ or whitespace character (space, tab, newline, carriage return, vertical tab or form feed), so such characters should not be used in string. Other characters (alphabetical, numerical and punctuations) are accepted. When nothing is left after truncation — an empty string, or one starting with ‘;’ or a blank character — vsearch issues a warning and writes a bare ;sample= annotation.

--sizeout

Add abundance annotations to sequence headers when writing fasta or fastq files. Add the pattern ;size=integer. Existing ;size= annotations are reported unchanged; entries without one receive ;size=1. For this command --sizein is not needed: abundance annotations are always parsed from the input headers.

--xee

Strip expected error (ee) annotations from sequence headers when writing fasta or fastq files. Search for the pattern [>@;]ee=float[;]. Expected error annotations are added by the synonymous options --fastq_eeout and --eeout described in vsearch-fastx_filter(1).

--xlength

Strip sequence length annotations from sequence headers when writing fasta or fastq files. Search for the pattern [>@;]length=integer[;]. Sequence length annotations are added by the --lengthout option.

--xsize

Strip abundance annotations from sequence headers when writing fasta or fastq files. Search for the pattern [>@;]size=integer[;]. Abundance annotations are added by the --sizeout option.

pairwise alignment options

These options modify the parameters of the pairwise alignment model. Modify with caution.

--gapext string: Set penalties for a gap extension. See vsearch-pairwise_alignment_parameters(7) for a complete description of the gap penalty declaration system. By default, the penalty is set to 2 for extending internal gaps and to 1 for extending terminal gaps, in both query and target sequences.
--gapopen string: Set penalties for a gap opening. See vsearch-pairwise_alignment_parameters(7) for a complete description of the gap penalty declaration system. By default, the penalty is set to 20 for opening internal gaps and to 2 for opening terminal gaps, in both query and target sequences.
--match integer: Set the score assigned to a match (i.e. equivalent nucleotides) in pairwise alignments. The default value is 2. Accepted values range from -32767 to +32767.
--mismatch integer: Set the score assigned to a mismatch (i.e. different nucleotides) in pairwise alignments. The default value is -4. Accepted values range from -32767 to +32767.

ignored options

--threads positive non-null integer: Command is not multithreaded, option has no effect (a warning is printed when more than one thread is requested).
--xn real number strictly greater than 1.0: Set the weight of ‘no’ votes, corresponding to the parameter beta in the chimera scoring function. Default value is 8.0. Increasing --xn reduces the likelihood of tagging a sequence as a chimera (less false positives, but also more false negatives). Decreasing --xn reduces false negative, but increases false positives.; Ignored by --chimeras_denovo: only the UCHIME scoring function reads this weight (a warning is printed when the option is given).

EXAMPLES

A simple way to filter out chimeras:

vsearch \
    --chimeras_denovo input.fasta \
    --quiet \
    --nonchimeras clean.fasta

Add option --tabbedout to log the sequences identified as chimeras, and option --log to record run parameters:

vsearch \
    --chimeras_denovo input.fasta \
    --quiet \
    --nonchimeras clean.fasta \
    --tabbedout chimeras.tsv \
    --log chimera_filtering.log

CITATION

Rognes T, Flouri T, Nichols B, Quince C, Mahé F. (2016) VSEARCH: a versatile open source tool for metagenomics. PeerJ 4:e2584 doi: 10.7717/peerj.2584

REPORTING BUGS

Submit suggestions and bug-reports at https://github.com/torognes/vsearch/issues, send a pull request on https://github.com/torognes/vsearch, or compose a friendly or curmudgeont e-mail to Torbjørn Rognes (torognes@ifi.uio.no).

AVAILABILITY

Source code and binaries are available at https://github.com/torognes/vsearch.

These manual pages are also published online at https://torognes.github.io/vsearch/.

COPYRIGHT

Contact: Torbjørn Rognes torognes@ifi.uio.no, Department of Informatics, University of Oslo, PO Box 1080 Blindern, NO-0316 Oslo, Norway

This software is dual-licensed and available under a choice of one of two licenses, either under the terms of the GNU General Public License version 3 or the BSD 2-Clause License.

GNU General Public License version 3

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.

The BSD 2-Clause License

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

ACKNOWLEDGMENTS

We would like to thank the authors of the following projects for making their source code available:

vsearch includes code from Google’s CityHash project by Geoff Pike and Jyrki Alakuijala, providing some excellent hash functions available under a MIT license.
vsearch includes code derived from Tatusov and Lipman’s DUST program that is in the public domain.
vsearch includes public domain code written by Alexander Peslyak for the MD5 message digest algorithm.
vsearch includes public domain code written by Steve Reid and others for the SHA1 message digest algorithm.
vsearch binaries may include code from the zlib library, copyright Jean-Loup Gailly and Mark Adler.
vsearch binaries may include code from the bzip2 library, copyright Julian R. Seward.