Option to turn OFF the filtering of low complexity regions

[ulad-litvin]

Good afternoon,

I am wondering if there is a possibility to turn OFF the filtering of low-complexity regions in DIGS or make the filtering optional (e.g. in the control file). I am searching for a protein with a region of lox complexity, and unfortunately, this region cannot be found unless the filtering is switched off. I tested it using DIGS, NCBI tblastn, and Ensembl tblastn. NCBI and Ensembl allow me to find the exon corresponding to this region only when I deselect the "Filter low complexity region" option, while I cannot find the region at all using DIGS. I hope it makes sense.

Thanks a lot in advance!

[robjgiff]

I think this should be possible. I will need to add an option to the command file to turn off filtering. If I understand correctly, this would mean supplying the following options to BLAST when running the forward BLAST search:

-dust no -soft_masking false

If you have any thoughts on this please let me know. I will do my best to implement this week.

[ulad-litvin]

I looked through the help pages for different blasts, and it looks like -dust filtering option exists only for blastn. Other blasts have -seg filtering option. According to the BLAST manual, segmasker is used to mask low-complexity regions of protein sequences, while dustmasker is used to do a similar thing for nucleotide sequences. I don't know for sure, but since the forward BLAST search uses tblastn, I think we need to change only one thing:

-seg no
# -soft_masking is false by default in tblastn

Let me know what you think.

[robjgiff]

Thanks this is helpful - It is possible to perform DIGS using blastn, so it looks as though I will need to implement this slightly differently depending on which BLAST program is being used.

[robjgiff]

I've updated the DIGS code that BLAST arguments can now be provided in the .ctl file, in the screensets block. e.g.

BEGIN SCREENSETS;
        query_na_fasta=/home2/rg128p/DIGS/projects/erv/fasta/herv/HERV_LTR-probes.fna
        reference_na_fasta=/home2/rg128p/DIGS/projects/erv/fasta/herv/HERV_LTR-references.fna;
        consolidated_reference_na_fasta=/home2/giff01r/DIGS/projects/erv/fasta/herv/HERV_LTR-references.fna;
        output_path=./tmp/;
        blast_bin_path='';
        bitscore_min_blastn=30;
        seq_length_minimum=50;
        defragment_range=50;
        consolidate_range=200;
        num_threads=8;
        dust=no;
        softmasking=false;
ENDBLOCK;

[ulad-litvin]

Hi Rob, I've been testing DIGS for the last couple of weeks. I've run it with the same settings as before, but this time turning off the pre-filtering.

BEGIN SCREENSETS;
    query_aa_fasta=/home2/2820395l/documents/digs_shisa5_test_4/digs_shisa5_probes_v1.fa;
    reference_aa_fasta=/home2/2820395l/documents/digs_shisa5_test_4/digs_shisa5_references_v1.fa;
    output_path=/home2/2820395l/documents/digs_shisa5_test_4/digs_shisa5_mammals_test_4_results/;
    bitscore_min_tblastn=41;
    seq_length_minimum=70;
    defragment_range=10;
    num_threads=2;
    seg=no;
    dust=no;
    softmasking=false;
ENDBLOCK;

I've managed to find missing Pro-rich protein regions this time. However, the problem now is that they have the wrong labelling (assigned_gene_name). I'm wondering if this has something to do with pre-filtering for the reverse BLAST run. When I turn off the pre-filtering, does it work only for the forward BLAST run or the reverse run as well? For instance, I've noticed that tblastn is now running with -seg no option (as it should be when pre-filtering is off). Thanks a lot!

[robjgiff]

OK thanks, this is good to know about, I'll take a look and get back to you.