App-Egaz

 view release on metacpan or  search on metacpan

lib/App/Egaz/Command/blastlink.pm  view on Meta::CPAN

    $desc .= <<'MARKDOWN';

* <infile> is reports produced by `egaz blastn`
* <infile> can't be stdin

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/blastmatch.pm  view on Meta::CPAN

* <infile> is reports produced by `egaz blastn`
* <infile> can't be stdin
* --perchr produces fake runlists, there might be overlaps, e.g. I:17221-25234,21428-25459

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/blastn.pm  view on Meta::CPAN

* <infile> should be a blocked fasta file (*.fas)
* <genome.fa> is a multi-sequence fasta file contains genome sequences
* `makeblastdb` and `blastn` should be in $PATH

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 2 ) {
        my $message = "This command need two input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/exactmatch.pm  view on Meta::CPAN

    $desc .= <<'MARKDOWN';

* <infile> can't be stdin
* `mummer` or `sparsemem` should be in $PATH

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 2 ) {
        my $message = "This command need two input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/fas2vcf.pm  view on Meta::CPAN

* Steps:
    1. split .fas to a temp dir by `fasops split`
    2. convert each fasta files to .vcf by `snp-sites`
    3. concat every .vcf files by `bcftools`

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 2 ) {
        my $message = "This command need two input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";

lib/App/Egaz/Command/lastz.pm  view on Meta::CPAN

* [`lastz` help](http://www.bx.psu.edu/~rsharris/lastz/README.lastz-1.04.15.html)

* [`--isself`](http://www.bx.psu.edu/~rsharris/lastz/README.lastz-1.04.15.html#ex_self)

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 2 ) {
        my $message = "This command need two input files/directories.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !( Path::Tiny::path($_)->is_file or Path::Tiny::path($_)->is_dir ) ) {

lib/App/Egaz/Command/lav2axt.pm  view on Meta::CPAN

    $desc .= ucfirst(abstract) . ".\n";
    $desc .= <<MARKDOWN;

* infile == stdin means reading from STDIN

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";

lib/App/Egaz/Command/lav2psl.pm  view on Meta::CPAN

    $desc .= ucfirst(abstract) . ".\n";
    $desc .= <<MARKDOWN;

* infile == stdin means reading from STDIN

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";

lib/App/Egaz/Command/lpcnam.pm  view on Meta::CPAN

    * Human19vsChimp3 use medium and 5000
    * loose is chicken/human linear gap costs
    * medium is mouse/human linear gap costs
* Default names of target and query in .maf are defined by basename of <path/target> and <path/query>

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 3 ) {
        my $message = "This command need three input files/directories.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !( Path::Tiny::path($_)->is_file or Path::Tiny::path($_)->is_dir ) ) {

lib/App/Egaz/Command/maskfasta.pm  view on Meta::CPAN

    return "egaz maskfasta [options] <infile> <runlist.yml>";
}

sub description {
    my $desc;
    $desc .= ucfirst(abstract) . ".\n";

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 2 ) {
        my $message = "This command need two input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";

lib/App/Egaz/Command/multiz.pm  view on Meta::CPAN

* <maf dirs> are directories containing multiple .maf or .maf.gz files
* `multiz` should be in $PATH
* Use a modified [`multiz`](https://github.com/wang-q/multiz) supports gzipped .maf files
* [Original `multiz`](https://www.bx.psu.edu/miller_lab/)

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} < 1 ) {
        my $message = "This command need one or more input directories.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !( Path::Tiny::path($_)->is_dir ) ) {

lib/App/Egaz/Command/normalize.pm  view on Meta::CPAN

* infile == stdin means reading from STDIN
* Start coordinates of output is 1-based
* Set --tlen and/or --qlen on partitioned sequences
* Ported from kentUtils src/hg/utils/automation/blastz-normalizeLav

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/partition.pm  view on Meta::CPAN

* Start coordinates of output is 1-based
* <infile> can't be stdin
* There is no --output option, outputs must be in the same directory of inputs
* Each fasta files should contain only one sequence. `faops split-name` can be use to do this.

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} < 1 ) {
        my $message = "This command need one or more input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/prepseq.pm  view on Meta::CPAN

    * each files should contain only 1 sequences and names of file and sequence should be matched
* Create chr.sizes via `faops size`
* Create chr.2bit via `faToTwoBit`
* `faops` and `faToTwoBit` should be in $PATH

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file/directory.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !( Path::Tiny::path($_)->is_file or Path::Tiny::path($_)->is_dir ) ) {

lib/App/Egaz/Command/raxml.pm  view on Meta::CPAN

* <infile> should be a blocked fasta file (*.fas)
* use `fasops concat` to create a temp relaxed phylip file
* `raxmlHPC` or `raxmlHPC-*` should be in $PATH
* `raxml` is very sensitive to CPU resources. Any competitions on cores will dramatically slow down it.

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} < 1 ) {
        my $message = "This command need one or more input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/repeatmasker.pm  view on Meta::CPAN

* `repeatmasker` should be in $PATH
* suffix of masked files is .fa, when --outdir is the same as infile:
    * infile.fa may be replaced
    * infile.fasta or infile.others will coexist with infile.fa

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} < 1 ) {
        my $message = "This command need one or more input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_file ) {

lib/App/Egaz/Command/template.pm  view on Meta::CPAN

  directories

* By default, `RAxML` is used to produce a phylotree. Turn on `--fasttree` to use FastTree, which is
  less accurate and doesn't support outgroups by itself

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} < 1 ) {
        my $message = "This command need one or more directories.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        if ( !Path::Tiny::path($_)->is_dir ) {