What to do when ApicoAP software fails to predict due to unresponsiveness of the SignalP Server?

ApicoAP uses the web service interface of SignalP 3.1 and when this server is down (which happens once in a while), ApicoAP tool gives the following error:

SignalP server is not responding, check out 'signalP_serverLog' file for details.

When this happens, there is really no way to fix it from your or my end, except to wait for it to get fixed by SignalP admins. In the mean time, I can suggest you to use the command line interface of ApicoAP with the "-s" option where you provide SP prediction information in a file. You can prepare this file by using the SignalP website (http://www.cbs.dtu.dk/services/SignalP-3.0/) which is still on right now. ApicoAP user manual has info on how to use the command line interface and how to prepare an SP file formatted properly. Here is an example showing you how to use this interface.

Files required for the example: trial.fasta and trial.spCleav

Content of trial.fasta:

>124808114
MKLTYYYYAFCIIKVIASLRLDSYKKQNVFFLKAFFKNVINIKKERKKNLVNSFFYVGRRNKIPIKYSNN
EFYNIHNENYDNTKYYNNNNNTNDYDKKDNNVGKRKYSRKNMLYMNSSEKDNFLNENILKKKSSEKEIEQ
SLTPLNMDWVKVMNLIYSSNDIDATTLAFNAAMSAVEKKGCLSTMLDLIGTMKSKNIKPDLVSYKLVLSL
CDKYHLVDTAEILFEEMIESDKINPNYEIYAIMISCYAKTGNGYKAIELFEKLRNDPFVEEMRSLNITNT
NDNKENSNDLQTSIIHNNMEDNNNNNNNNDNNIYDDKFKHISNKIKNVENCSGKIQYSEYANVIYACNIS
NLYEQGIKYFEELLKSGKYMPSIFVFENIFDLLSKNGDYEKSLEYYNNLKNDPNFKKNINVNILNNLLKA
LSIHNKINVAEDIWNNEFDELLLTPNNLSYQILLKIYSHIDNYEKAFKLFKEMQVNKLLNNKNILPFIYT
IESTKNCGIYNYAIYVLRVAKLLNFKANDLLMLYNNTMISCINSKKYDVIISLYAELINMQQKDTSFQIN
INTLTFVLLAFKELNMKQDFINLKNIIIQRNYKLPPLCSKIFSETENY

Content of trial.spCleav:

124808114   18

Explanation

trial.fasta has the sequence info along with an identification number of a Plasmodium falciparum protein. If you input this sequence into SignalP 3.0 (copy the whole fasta file content and paste into the box at http://www.cbs.dtu.dk/services/SignalP-3.0/ ), select Eukaryotes, Both methods, No graphics and Standard options and you'll get the following:

>124808114
SignalP-NN result:
>124808114             length = 70
# Measure  Position  Value  Cutoff  signal peptide?
  max. C    19       0.351   0.32   YES
  max. Y    19       0.390   0.33   YES
  max. S     8       0.733   0.87   NO
  mean S     1-18    0.526   0.48   YES
       D     1-18    *0.458*   0.43   YES
# Most likely cleavage site between pos. *18* and 19: IAS-LR
SignalP-HMM result:
>124808114
Prediction: Non-secretory protein
Signal peptide probability: 0.016
Signal anchor probability: 0.000
Max cleavage site probability: *0.011* between pos. 18 and 19

Standard rule to decide if there is SP: D-score > 0.43 or HMM probability > 0.5. This paper uses tighter thresholds of 0.48 and 0.9, respectively. PlasmoDB uses threshold 0.5 on both and also checks out S-score, C-score, Y-score, Mean-S decisions, if 3 out of 4 is yes, then considers there is SP. Many different rules out there:)

We used standard rules in ApicoAP v2, and tighter rules in ApicoAP v3. In the above example numbers to check against the threshold and the SP cleavage site info we need is marked with star(*).

Required files (given above) are fasta and SP cleavage files in correct format for this example. Following command run will generate out.txt if the trial.spCleav and the trial.fasta files are placed in the same directory as ApicoAP_CL.py (which is under wherever you opened ApicoAP_05_20_13.zip)

M:\GokcenArchive\GokcenDissertation\ApicoAP_05_20_13>python ApicoAP_CL.py -m PFA
L -s trial.spCleav -f trial.fasta -o out.txt

ApicoAP predictions are successfully generated, check out the output file!