filter.seqs

filter.seqs removes columns from alignments based on a criteria defined by the user. For example, alignments generated against reference alignments (e.g. from RDP, SILVA, or greengenes) often have columns where every character is either a ‘.’ or a ‘-‘. These columns are not included in calculating distances because they have no information in them. By removing these columns, the calculation of a large number of distances is accelerated. Also, people also like to mask their sequences to remove variable regions using a soft or hard mask (e.g. Lane’s mask). This type of masking is only encouraged for deep-level phylogenetic analysis, not fine level analysis such as that needed with calculating OTUs. This tutorial uses the data files in amazondata.zip.

Default settings

To run filter.seqs you need to provide your sequences to be filtered in either fasta, nexus, clustal, or phylip format. The output will be in fasta format. By default, any column with a ‘-‘ in every sequence is removed from the alignment and put into a *.filter.fasta file. For example the sequences in the file amazon.unique.align are the sequences in amazon.unique.fasta aligned to the greengenes alignment. The sequences in amazon.unique.fasta are generally less than 1 kb, but vary considerably in length. We’d like to only consider those columns in the alignments that overlap. Also, considering the overall alignment length is 7682 characters, we can probably remove a lot of positions from the alignment to speed up our distance calculations. Try the following command:

mothur > filter.seqs(fasta=amazon.unique.align)

The resulting alignment in the amazon.unique.filter.fasta file will be 2305 characters long and the average unaligned sequence is about 400 bp long. The file, amazon.unique.filter, is a series of 0’s and 1’s that indicate, which columns were included in the filtered alignment. This file can be used with the hard option below. Also, a series of statistics are written to the screen that indicate how many columns were used in generating the filter and the number of sequences that the filter was based on:

Length of filtered alignment: 1898
Number of columns removed: 5784
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

You can also enter multiple files to filter together. You do this by separating the fasta files with ‘|’ characters. mothur creates a filter and filters the sequences as if they were all in the same file, but outputs separate .filtered.fasta files.

mothur > filter.seqs(fasta=amazon.unique.align|core_set_aligned.imputed.fasta)

Options

vertical

By default vertical option is set to T, and any column that only contains gap characters (i.e. ‘-‘ or ‘.’) is ignored.

mothur > filter.seqs(fasta=amazon.unique.align, vertical=T)

This can be turned off by setting vertical to F:

mothur > filter.seqs(fasta=amazon.unique.align, vertical=F)

Length of filtered alignment: 7682
Number of columns removed: 0
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

Note that nothing was removed from the alignment.

trump

The trump option will remove a column if the trump character is found at that position in any sequence of the alignment. You can use any character with the trump setting (‘.’, ‘-‘, ‘N’, etc). NOTE: having one or two sequences included that don’t align with the bulk of your sequences may lead to all columns being removed by the trump option!

The align.seqs, NAST, and SILVA aligners will precede the first base of the sequence with a string of periods and the columns from the last base to the end of the alignment will also be a string of periods. You may want to remove these columns:

mothur > filter.seqs(fasta=amazon.unique.align, trump=., vertical=F)

Length of filtered alignment: 2305
Number of columns removed: 5377
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

Sometimes for protein coding sequences people like to remove any column that contains a gap in any sequence. This can be achieved with the command:

mothur > filter.seqs(fasta=amazon.unique.align, trump=-, vertical=F)

Length of filtered alignment: 2030
Number of columns removed: 5652
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

This does nothing:

mothur > filter.seqs(fasta=amazon.unique.align, trump=,  vertical=F)

Length of filtered alignment: 7682
Number of columns removed: 0
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

It is suggested that the trump and vertical options be used together for maximum speed up of downstream distance calculations:

mothur > filter.seqs(fasta=amazon.unique.align, trump=., vertical=T)

or

mothur > filter.seqs(fasta=amazon.unique.align, trump=.)

Because ‘vertical=T’ is the default, both commands will produce the following output:

Length of filtered alignment: 577
Number of columns removed: 7105
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

Thus, the final alignment is only 577 columns long. This is more than a 13-fold reduction in the alignment length and will result in a speed up of more than 13-fold in the distance calculation. Notice:In 1.20 vesion, the result of “filter.seqs(fasta=amazon.unique.align, trump=.)” is same of “filter.seqs(fasta=amazon.unique.align, trump=., vertical=F)”

soft

A soft mask removes any column where the dominant base (i.e. A, T, G, C, or U) does not occur in at least a designated percentage of sequences. For resolution of deep branching in phylogenetics, it is common to require that the dominant character occur in at least 50% of the sequences. Such a filter can be done in mothur as follows:

mothur > filter.seqs(fasta=amazon.unique.align, soft=50, vertical=F)

Length of filtered alignment: 440
Number of columns removed: 7242
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

The value given to soft must be an integer between 0 and 100. Again, this filter option is only suggested for resolving the relationships between deep branches in the phylogenetic trees, not for calculating OTUs.

hard

Use of the file option will allow one to apply a hard mask to the sequences (e.g. the lane mask). The inputted file should only contain one line consisting of 0’s and 1’s. A “0” indicates that the column should be excluded and a “1” indicates that the column should be included. To use a hard mask you would enter:

mothur > filter.seqs(fasta=amazon.unique.align, hard=Lane1349.gg.filter, vertical=F)

Length of filtered alignment: 1232
Number of columns removed: 6450
Length of the original alignment: 7682
Number of sequences used to construct filter: 96

Again, this filter option is only suggested for resolving the relationships between deep branches in the phylogenetic trees, not for calculating OTUs.

processors

The processors parameter allows you to run the command with multiple processors. Default processors=Autodetect number of available processors and use all available.

mothur > filter.seqs(fasta=amazon.unique.align, processors=2)

Revisions

  • 1.24.0 - Paralellized for Windows.
  • 1.25.0 - When run with “current”, it generated “.filter” and not “filename.filter” file.
  • 1.40.0 - Rewrite of threaded code. Default processors=Autodetect number of available processors and use all available.
  • 1.44.0 - Changes filter.seqs fasta file deliminator from ‘-‘ to ‘|’ to allow for ‘-‘’s in filenames.