obiclean
: tags a set of sequences for PCR/sequencing errors identification¶
obiclean
is a command that classifies sequence records either as head
, internal
or singleton
.
- For that purpose, two pieces of information are used:
- sequence record counts
- sequence similarities
- S1 a sequence record is considered as a variant of S2 another sequence record if and only if:
count
of S1 divided bycount
of S2 is lesser than the ratio R. R default value is set to 1, and can be adjusted between 0 and 1 with the-r
option.- both sequences are related to one another (they can align with some differences,
the maximum number of differences can be specified by the
-d
option).
- Considering S a sequence record, the following properties hold for S tagged as:
head
:- there exists at least one sequence record in the dataset that is a variant of S
- there exists no sequence record in the dataset such that S is a variant of this sequence record
internal
:- there exists at least one sequence record in the dataset such that S is a variant of this sequence record
singleton
:- there exists no sequence record in the dataset that is a variant of S
- there exists no sequence record in the dataset such that S is a variant of this sequence record
By default, tagging is done once for the whole dataset, but it can also be done sample by sample
by specifying the -s
option. In such a case, the counts are extracted from the sample
information.
Finally, each sequence record is annotated with three new attributes head
, internal
and
singleton
. The attribute values are the numbers of samples in which the sequence record has
been classified in this manner.
obiclean
specific options¶
-
-d
<INTEGER>
,
--distance
=<INTEGER>
¶ Maximum numbers of differences between two variant sequences (default: 1).
-
-s
<KEY>
,
--sample
=<KEY>
¶ Attribute containing sample descriptions.
-
-r
<FLOAT>
,
--ratio
=<FLOAT>
¶ Threshold ratio between counts (rare/abundant counts) of two sequence records so that the less abundant one is a variant of the more abundant (default: 1, i.e. all less abundant sequences are variants).
-
-C
,
--cluster
¶
Switch
obiclean
into its clustering mode. This adds information to each sequence about the true.
-
-H
,
--head
¶
Select only sequences with the head status in a least one sample.
-
-g
,
--graph
¶
Creates a file containing the set of DAG used by the obiclean clustering algorithm. The graph file follows the dot format
Options to specify input format¶
Restrict the analysis to a sub-part of the input file¶
-
--skip
<N>
¶ The N first sequence records of the file are discarded from the analysis and not reported to the output file
-
--only
<N>
¶ Only the N next sequence records of the file are analyzed. The following sequences in the file are neither analyzed, neither reported to the output file. This option can be used conjointly with the –skip option.