Marco Necci, Damiano Piovesan and Silvio C.E. Tosatto
Version 3.10.0
Docker container: https://github.com/BioComputingUP/MobiDB-lite_docker
MobiDB-lite executes up to 9 different disorder predictors, collects the outputs and calculates a consensus. The consensus is generated by measuring predictors agreement. At least 62.5% of predictors must agree to assign disorder state to a residue. Then a mathematical morphology (MM) dilation/erosion processing is applied. Finally short regions are filtered out.
MobiDB-lite works both for 32-bit and 64-bit architectures in Linux system. It requires Java (version >=7.x) and Python (version 3.x.x). By default it uses 64-bit executables, but 32-bit version can be called by a command line option (see below).
The following command prints the help page on screen::
python3 mobidb_lite.py -h
By default, MobiDB-lite searches the binx/ folder (that includes predictors
executables) in its own directory. So, MobiDB-lite can be executed by just
providing a multi-fasta input:
python3 mobidb_lite.py /examples/multi.fasta
MobiDB-lite is able to use predictors executables from other paths. See Configuration section for further details.
By default, MobiDB-lite runs on single thread but it can exploits up to 7 threads. When less than 7 CPUs are available, the "-t" option allows to calibrate the load:
python3 mobidb_lite.py examples/multi.fasta -t 4
MobiBD-lite is designed to work out of the box. For a basic usage you should not need to configure anything. Nonetheless, MobiDB-lite offers some configuration possibilities.
MobiDB-lite is able to use predictors executables from other paths. Paths of
single predictors can be configured in the config.ini or custom configuration file.
You should always use absolute paths when configuring binx directories.
Notice that when paths ar not set in this file, MobiDB-lite will look for the
executables in the binx/ folder.
Most predictors' output is a score ranging from 0 to 1. To transform into a binary
prediction a threshold is applied. Thresholds are usually defined in publications
relative to single predictors. Predefined thresholds are obtained from such publications.
However, thresholds of single predictors can be configured in the config.ini or
custom configuration file.
Log file and log level can be configured through command line arguments, --log and
--logLevel respectively. Default log level is set to ERROR.
The output is printed on screen by default. To save it on a file the "-o" option has to be used:
python3 mobidb_lite.py examples/multi.fasta -o out_file.txt
The output is provided in 3 different formats. The output format can be chosen
with the -f option, which accepts one of the following values: interpro.
fasta, caid, mobidb4. The default short version
(option interpro) includes the protein name and start/end position of the
disorder regions, one region per line. All elements are separated by a TAB:
sp|Q15648|MED1_HUMAN 609 706
sp|Q15648|MED1_HUMAN 652 681 Polar
sp|Q15648|MED1_HUMAN 792 820
sp|Q15648|MED1_HUMAN 806 820 Polar
sp|Q15648|MED1_HUMAN 874 893
sp|Q15648|MED1_HUMAN 947 1566
sp|Q15648|MED1_HUMAN 995 1020 Polyampholyte
sp|Q15648|MED1_HUMAN 1078 1156 Low complexity
sp|Q15648|MED1_HUMAN 1158 1197 Polar
sp|Q15648|MED1_HUMAN 1219 1292 Low complexity
sp|Q15648|MED1_HUMAN 1328 1348 Polar
sp|Q15648|MED1_HUMAN 1349 1366 Polyampholyte
sp|Q15648|MED1_HUMAN 1367 1382 Low complexity
sp|Q15648|MED1_HUMAN 1421 1484 Polar
sp|Q15648|MED1_HUMAN 1508 1530 Polyampholyte
sp|Q15648|MED1_HUMAN 1531 1556 Polar
Sequence features can be silenced by using option -sf, inactive by default. When using this option, sequence features are skipped:
sp|Q15648|MED1_HUMAN 609 706
sp|Q15648|MED1_HUMAN 792 820
sp|Q15648|MED1_HUMAN 874 893
sp|Q15648|MED1_HUMAN 947 1566
Option fasta gives the output in fasta format (new in version 3.8.2).
In this format D is disorder, S is structure and region subtypes are
given as figures from 1 to 8:
>sp|Q15648|MED1_HUMAN
SSS...SSDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
D888888888888888888888888888888DDDDDDDDDDDDDDDDDDD
DDDDDDSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSDDDDDDDDD
DDDDD888888888888888SSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSDDDDDDDDDDDDDDDDDDDDSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD111111
11111111111111111111DDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDD77777777777777777777777
77777777777777777777777777777777777777777777777777
777777D8888888888888888888888888888888888888888DDD
DDD...
Sequence features can be silenced by using option -sf, inactive by
default. When using this option, sequence features are expressed as D:
>sp|QDDDDD|MEDD_HUMAN
SSS...SSDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSDDDDDDDDDDDDDDDDDDDDSSSSSSS
SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
DDD...
Option caid gives the output with one residue per line (new in version 3.8.4).
In this format each line will will have four values: pos res score state. Where
pos is the residue position in the sequence (starting from 1), res
is the amino acid one letter code, score is the agreement score of the consensus,
state if the predictor decision between order (0) and disorder (1):
>sp|Q15648|MED1_HUMAN mobidb_lite
1 M 0.75 0
2 K 0.75 0
... ... ... ...
16 M 0.49 0
17 S 0.571 1
18 S 0.635 1
Option mobidb4 provides an extended output and structured output that is used
to create annotations for MobiDB4 entries:
{
"acc": "sp|Q15648|MED1_HUMAN",
"sequence": "MKAQASQAL...",
"length": 1581,
"prediction-disorder-mobidb_lite": {
"regions": [[609, 706], [792, 820], [874, 893], [947, 1566]],
"scores": [0.75, 0.75, 0.75, 0.875, 0.75, 0.875, ... ],
"content_count": 767,
"content_fraction": 0.485
},
"prediction-disorder-th_50": {...},
"prediction-low_complexity-merge": {...},
"prediction-disorder-iupl": {...},
"prediction-disorder-iups": {...},
"prediction-disorder-espN": {...},
"prediction-disorder-espD": {...},
"prediction-disorder-espX": {...},
"prediction-disorder-glo": {...},
"prediction-disorder-dis465": {...},
"prediction-disorder-disHL": {...},
"prediction-disorder-vsl": {...},
"prediction-low_complexity-seg": {...},
"prediction-low_complexity-pfilt": {...},
"prediction-helix-fess": {...},
"prediction-sheet-fess": {...},
"prediction-coil-fess": {...},
"prediction-rigidity-dynamine": {...},
"prediction-lip-anchor": {...}
}
If you are not seeing any output try to use the -fc option to force output in case of single predictors failures.
MobiDB-lite won't work on OS-X since some of the underlying predictors are not compiled for it.
If you encounter any bug or strange behaviour, please don't hesitate and contact us at biocomp@bio.unipd.it.