HMM PD00134
Back to HMM Home
Why build in-house profile for GRAM
The GRAM domain in MTMR2 is from 74 to 185 as shown in crystal structure [1]. It has 7 beta sheets and 1 alpha helix, resembling Pleckstrin (PH) domain. The profiles in Pfam and SMART database are not able to capture the complete domain: Pfam gives 78-138 (envelope 68-139); SMART gives 71-139. The region covers beta sheets 1-5, but not beta sheet 6, 7 and alpha helix 1. We therefore built a profile to capture the full GRAM domain. Below are the sequence of GRAM domain of human MTMR2 determined by structure with 5 flanking residues at both N- and C-terminal (69-190):
KLAEMEEPPLLPGENIKDMAKDVTYICPFTGAVRGTLTVTNYRLYFKSMERDPPFVLDASLGVINRVEKIGGASSRGENSYGLETVCKDIRNLRFAHKPEGRTRRSIFENLMKYAFPVSNNL
Note: We tried to align the full sequences of myotubularins in phosphatome.net database. The region of GRAM domain defined by MTMR2 GRAM (74-185) does not seem well aligned. Instead, we PSI-BLASTed the GRAM domain sequences of individual subfamilies.
Here is the list of GRAM HMMs:
- Pfam GRAM
- SMART GRAM
- Custom GRAM
- GRAM-MTMR_1: MTMR1, MTMR5, MTMR6, MTMR9 and MTMR10 subfamilies
- GRAM-MTMR3: MTMR3 subfamily
- GRAM-MTMR14: MTMR14 subfamily
- GRAM-MTMR: myotubularin family
How we built the HMM
We found the myotubularin GRAM domains falls into three groups in sequence similarity when we performed PSI-BLAST. We therefore built subfamily-specific HMMs for each of the three groups, and combined them into one HMM.
Group 1: MTMR1, MTMR5, MTMR6, MTMR9 and MTMR10 subfamilies
When we PSI-BLASTed the GRAM domain sequence of MTMR10 defined by CDD profile PH-GRAM_MTMR10 (see below) against SWISS-PROT dataset, we found the hits of MTMR1, MTMR5, MTMR6, MTMR9 and MTMR10 subfamilies and converged at 7th round. We selected the sequences with coverage higher than 80% of the query sequence (see below). We then downloaded the aligned sequences, performed multiple sequence alignment using Clustal Omega and built HMM profile using HMM3b.
TDDKINSEPKIKKLEPVLLPGEIVVNEVNFVRKCIATDTSQYDLWGKLICSNFKISFITDDPMPLQKFHYKNLLLGEHDVPLTCIEQIVTVNDHKRKQKVLGPNQKLKFNPTELIIYCKDFRVVRFRFDESGPESAKKVCLAIAHYSQPTDLQLLFAFEYVGKKYHNSVSSINGM
Group 2: MTMR14 subfamily
We PSI-BLASTed the GRAM domain sequence of MTMR14 defined by CDD profile PH-GRAM_MTMR14 against SWISS-PROT dataset (see below). There was few hit, so we searched against RefSeq dataset, instead. The search converged at 3rd round and the hits were only from MTMR14. We selected the sequences with coverage higher than 80% of the query sequence (see below). We then downloaded the aligned sequences, remove redundant sequences (by CD-HIT, threshold 0.9) and performed multiple sequence alignment using Clustal Omega and built HMM profile using HMM3b.
RFPFPVIYFHNKNLCRSSTLSKKIEYMFQSGVNSMKKQFstapttnqspnpnpnnnttivnIQPQAEETEIGdQNMENLRnNDINAIKHLSVKYICDLMVEnKKKKFGFYvcSSEKADMHDRYTKQFV
Group 3: MTMR3 subfamily
We PSI-BLASTed the GRAM domain sequence of MTMR3 defined by CDD profile PH-GRAM_MTMR3 against SWISS-PROT dataset (see below). There was few hit, so we searched against RefSeq dataset, instead. The search converged at 2nd round and the hits were only from MTMR3. We selected the sequences with coverage higher than 80% of the query sequence (see below). We then downloaded the aligned sequences, remove redundant sequences (by CD-HIT, threshold 0.9) and performed multiple sequence alignment using Clustal Omega and built HMM profile using HMM3b.
MDEETRHSLECIQANQIFPRKQLIREDENLQVPFLELHGESTEFVGRAEDAIIALSNYRLHIKFKESLVNVPLQLIESVECRDIFQLHLTCKDCKVIRCQFSTFEQCQEWLKRLNNAIRPPAKIEDLFSFAYHAWCMEVYASEKEQHG
Combine the HMMs into one
We first aligned the alignments of the three groups by program Clustal Omega using profile-vs-profile mode. However, the quality of the alignment is not well enough for manual adjustment. We aligned the sequences from the three groups by PROMALS3D followed by manual adjustment, and built the HMM.
