1\section{Profile construction steps, in detail} 2\label{section:pipeline} 3\setcounter{footnote}{0} 4 5\begin{description} 6 7\item [\textbf{Read and validate input alignment}] 8 9\item [\textbf{Calculate relative sequence weights}] 10 11\item [\textbf{Define sequence fragments}] 12 13\item [\textbf{Determine model architecture}] 14 15\item [\textbf{Calculate effective sequence number}] 16 17\item [\textbf{Parameterize the model}] 18 19\item [\textbf{Transfer annotation from MSA to model}] 20 21\item [\textbf{Calibrate the model}] 22 23\item [\textbf{Create the ``post'' MSA}] 24 25\end{description} 26 27In more detail, each step is described in the following 28subsections.\footnote{To study the relevant source code, start with 29 \ccode{p7\_builder.c::p7\_Builder()}.} 30 31 32\subsection{Alignment input and validation} 33 34HMMER is designed to work on input 35 36 37\subsection{Relative sequence weighting} 38 39\subsection{Sequence fragment definition} 40 41\subsection{Model architecture determination} 42 43\subsection{Effective sequence number (absolute sequence weights)} 44 45\subsection{Model parameterization} 46 47\subsection{Annotation transfer from MSA to model} 48 49\subsection{Model calibration} 50 51\subsection{Create the ``post'' MSA} 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67