GapMind for catabolism of small carbon sources

 

Alignments for a candidate for liuA in Algoriphagus aquaeductus T4

Align Isovaleryl-CoA dehydrogenase (EC 1.3.8.4) (characterized)
to candidate WP_111393797.1 CLV31_RS14820 acyl-CoA dehydrogenase

Query= reanno::psRCH2:GFF1051
         (387 letters)



>NCBI__GCF_003253485.1:WP_111393797.1
          Length = 379

 Score =  298 bits (763), Expect = 2e-85
 Identities = 156/375 (41%), Positives = 231/375 (61%)

Query: 6   LNFALGETIDMLREQVQAFVAAEIAPRAEAIDQENLFPADMWRKFGEMGLLGVTVSEEYG 65
           +NF L E    ++E  + F  +E+ P     D E  FP +  +K G++G LG+ V  +Y 
Sbjct: 1   MNFQLTEEHLAVQEAAREFAQSELLPGVIDRDSEAKFPHEQIKKMGDLGFLGMMVDPKYN 60

Query: 66  GAGLGYLAHVVAMEEISRGSASVALSYGAHSNLCVNQINRNGNPEQKARYLPKLISGEHV 125
           G G+  +++V+AMEE+S+  AS ++S   +++L    + + G+ EQK +YL +L +GE +
Sbjct: 61  GGGMDTISYVIAMEELSKIDASASVSMSVNNSLVCWGLEKYGSEEQKQKYLTRLATGEIL 120

Query: 126 GALAMSEPNAGSDVVSMKLRAEKRGDRYVLNGSKTWITNGPDANTYVIYAKTDLDKGAHG 185
           GA  +SEP AGSD  S +  AE  GD YVLNG+K WITNG  A+ Y++ A+TD  KG  G
Sbjct: 121 GAFCLSEPEAGSDATSQRTSAEWNGDHYVLNGTKNWITNGSTASVYLVIAQTDASKGHKG 180

Query: 186 ITAFIVERDWKGFSRGNKFDKLGMRGSNTCELFFDDVEVPQENVLGAENGGVKVLMSGLD 245
           I+ F+VE+ W+GF  G K DKLG+RGS+T  L F DV+VP EN +G E  G    M  L+
Sbjct: 181 ISVFLVEKGWEGFVIGKKEDKLGIRGSDTHSLMFTDVKVPAENRIGEEGFGFTFAMETLN 240

Query: 246 YERVVLAGGPTGIMQSCLDVVVPYIHDRKQFGQSIGEFQFIQGKVADMYTQLNASRAYLY 305
             R+ +A    GI     ++ + Y  +RK FG+ I + Q IQ K+ADM TQ+ A+R  +Y
Sbjct: 241 GGRIGIAAQALGIASGAYELALAYSKERKAFGKPISQHQAIQFKLADMATQIEAARLLVY 300

Query: 306 AVAQACDRGETTRKDAAGVILYTAENATQMALQAIQILGGNGYINEFPTGRLLRDAKLYE 365
             A   D+GE     +A   LY +E A  + ++A+Q+ GG GY+ E+   RL+RDAK+ +
Sbjct: 301 KAAWLKDQGEDYAHASAMAKLYASEVAMNVTVEAVQVHGGYGYVKEYHVERLMRDAKITQ 360

Query: 366 IGAGTSEIRRMLIGR 380
           I  GTSEI+R++I R
Sbjct: 361 IYEGTSEIQRIVISR 375


Lambda     K      H
   0.318    0.135    0.391 

Gapped
Lambda     K      H
   0.267   0.0410    0.140 


Matrix: BLOSUM62
Gap Penalties: Existence: 11, Extension: 1
Number of Sequences: 1
Number of Hits to DB: 380
Number of extensions: 15
Number of successful extensions: 1
Number of sequences better than 1.0e-02: 1
Number of HSP's gapped: 1
Number of HSP's successfully gapped: 1
Length of query: 387
Length of database: 379
Length adjustment: 30
Effective length of query: 357
Effective length of database: 349
Effective search space:   124593
Effective search space used:   124593
Neighboring words threshold: 11
Window for multiple hits: 40
X1: 16 ( 7.3 bits)
X2: 38 (14.6 bits)
X3: 64 (24.7 bits)
S1: 41 (21.7 bits)
S2: 50 (23.9 bits)

This GapMind analysis is from Sep 24 2021. The underlying query database was built on Sep 17 2021.

Links

Downloads

Related tools

About GapMind

Each pathway is defined by a set of rules based on individual steps or genes. Candidates for each step are identified by using ublast (a fast alternative to protein BLAST) against a database of manually-curated proteins (most of which are experimentally characterized) or by using HMMer with enzyme models (usually from TIGRFam). Ublast hits may be split across two different proteins.

A candidate for a step is "high confidence" if either:

where "other" refers to the best ublast hit to a sequence that is not annotated as performing this step (and is not "ignored").

Otherwise, a candidate is "medium confidence" if either:

Other blast hits with at least 50% coverage are "low confidence."

Steps with no high- or medium-confidence candidates may be considered "gaps." For the typical bacterium that can make all 20 amino acids, there are 1-2 gaps in amino acid biosynthesis pathways. For diverse bacteria and archaea that can utilize a carbon source, there is a complete high-confidence catabolic pathway (including a transporter) just 38% of the time, and there is a complete medium-confidence pathway 63% of the time. Gaps may be due to:

GapMind relies on the predicted proteins in the genome and does not search the six-frame translation. In most cases, you can search the six-frame translation by clicking on links to Curated BLAST for each step definition (in the per-step page).

For more information, see:

If you notice any errors or omissions in the step descriptions, or any questionable results, please let us know

by Morgan Price, Arkin group, Lawrence Berkeley National Laboratory