tblastn cuts querry even when alligment cut be better if extended

Question

[john]

I am trying to blast proteins on a single genome. So I build a nucleotid database of the genome and used tblastn to align the query proteins on the genome. I did not change any standard parameters.

tblastn -query protein.fasta -db genome.fasta

The protein.fasta file looks like this:

>EDR98569.1 putative toxin-antitoxin system, toxin component [Anaerostipes caccae DSM 14662]
MDLNYHDILVKILDVYRDCEVQSFPIDCYSILRHYGYRIFTYQNIRDINERLYQYCRNYS
EDAFRYGAKRIIAYDENKSPFRIRFSIMHELGHIMLGHSRECAYNEQQANFFASNILAPR
MAIHFAQCRNEDDVSSVFQISREAGSYAFQNYRLWKESAAREVSDVDEAMYRHFYHDERE
EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRMLLQIEQQQLNNL

The alligment looks like this:

Query  1      MDLNYHDILVKILDVYRDCEVQSFPIDCYSILRHYGYRIFTYQNIRDINERLYQYCRNYS  60
              +DLNYHDILVKILDVYRDCEVQSFPIDCYSILRHYGYRIFTYQNIRDINERLYQYCRNYS
Sbjct  37493  VDLNYHDILVKILDVYRDCEVQSFPIDCYSILRHYGYRIFTYQNIRDINERLYQYCRNYS  37672

Query  61     EDAFRYGAKRIIAYDENKSPFRIRFSIMHELGHIMLGHSRECAYNEQQANFFASNILAPR  120
              EDAFRYGAKRIIAYDENKSPFRIRFSIMHELGHIMLGHSRECAYNEQQANFFASNILAPR
Sbjct  37673  EDAFRYGAKRIIAYDENKSPFRIRFSIMHELGHIMLGHSRECAYNEQQANFFASNILAPR  37852

Query  121    MAIHFAQCRNEDDVSSVFQISREAGSYAFQNYRLWKESAAREVSDVDEAMYRHFYHDERE  180
              MAIHFAQCRNEDDVSSVFQISREAGSYAFQNYRLWKESAAREVSDVDEAMYRHFYHDERE
Sbjct  37853  MAIHFAQCRNEDDVSSVFQISREAGSYAFQNYRLWKESAAREVSDVDEAMYRHFYHDERE  38032

Query  181    EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRM  227
              EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRM
Sbjct  38033  EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRM  38173

The protein has an alternative start codon as you can see. And the querry is shortened by the last 13 AS

LLQIEQQQLNNL

So far so good this can be the expected behavior if not further matches could be made when extending the query. But if I get the genome: ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/154/305/GCA_000154305.1_ASM15430v1 and look what is downstream from the alligment I get this sequence:

VDLNYHDILVKILDVYRDCEVQSFPIDCYSILRHYGYRIFTYQNIRDINERLYQYCRNYSEDAFRYGAKRIIAYDENKSPFRIRFSIMHELGHIMLGHSRECAYNEQQANFFASNILAPRMAIHFAQCRNEDDVSSVFQISREAGSYAFQNYRLWKESAAREVSDVDEAMYRHFYHDEREEFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRMLLQIEQQQLNNL*EH

As you can see blast could have extended the query and match all the 13 AS, hence the score would have improved. So my question is now:

Why is tblastn show this behavior?

How can I force tblastn to align the full querry? (-qcov_hsp_perc 100 is not an option as this just filters the output)

If you wanne try for your self you can use the webservice https://blast.ncbi.nlm.nih.gov/Blast.cgi and under Database the wgs db and use the taxid 411490. The result is the same.

Oh and has nothing to do with a specific length of the query. For example this protein is much longer and also is cut when aligned by 6 AS.

>EDR97547.1 repeat protein [Anaerostipes caccae DSM 14662]
MGGGLSRSENTENITETFLGADETIVPLVNKVSEKSLKKTKEGQTMLDVSLGNIRINQTG
ASGGGLSEAETELNPNGYYITGKTKSYNVVVAKGVKTDLIFDSVEIESNSTTQSCVIVSH
ADVTITLKGVNIWGCNYGTSHDTNGGAVLAKNGMDGFLTVQCEYADQEGHLCDDNCSTLI
ATGNVVHAGAIGSTISNVTTASECGFCNFRVKGGNLEVSGGTHVAGIGSACNSQVYAGGY
TKNIYISGGNIKATGTERGPGIGGGYGSDFDGLYITGGKVEARGGASAPGIGTSSGQDGT
YKLKNVHISGGDTIVIAIGDKSTKMPGIGSAYGNANVSNVTAEPDPGYQGYIQDGTSLED
YSFMEGTPFHEKTDIRVGRFYTKVYFGPFRDVNGIEDDTKEQIGANHVISKSGGKPFTEE
LLHHLTKVTGKQENGTNFPPEQLTLADLSELETINAAKTKGEIGDFPLTYTTPNGTKATV
TVYLRNDGEDAGGFDKENIKEQIGANDFTKETGGSPFTEEDIKHLGEVKGKGKEGSNISL
DDFSVDQEQFKKINEHKTQGKAGEFELTYSDAKGNKVTVTVTLAGEYDAITENPDTGEMI
KGKHIISKTGGDGFTKEQLKGLSMVKAVDKDGTEIPTEDLSFAEEEQIAAINAAKTAGKT
GDFPLTFRTPDNTTVTITVFLRDEGTDAAKEGQDDPFSVIGANHTIQPTKGEPFTEEQII
DLCQAKGKDKNLDNAKILVDESQLAVINKAKKDGKTGVFDLTFSLSDGNKATVKVTLTGD
HRVSFDPDGGDYQPETQTVKGGDCAEAPRDPAKEGYVFEGWYYIDEDGNEVKWDFKTPVH
SDVKLKAKWKEADRTETTAVPTTAKKPKQKKTVPEWEYKKRVRRKRVSRTGDERQILCLI
VMFGAALTGLAAGIRKKRR

Answer 1

[lieven.sterck]

Ok, confirmed the behavior.

To resolve it you should deactivate the low complexity option. As I expected the last part is largely being masked and thus does not end up in the alignment. Keep in mind though that this might also lead to (more) false positives.

Query  181    EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRMLLQIEQQQLNNL  239
              EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRMLLQIEQQQLNNL
Sbjct  38033  EFIYSIKPCMICGETIYNSSEDLCLHCRMEHIRRQHTPLYTSRNDRMLLQIEQQQLNNL  38209

And a little warning on top:

(-qcov_hsp_perc 100 is not an option as this just filters the output)

I would not jump to that conclusion (==that it only "filters" the output) to easily, as a number of trending threads here on biostars will indicate