Skip to content

Commit

Permalink
review edits
Browse files Browse the repository at this point in the history
  • Loading branch information
davidbenjamin committed Mar 25, 2019
1 parent 05ab4eb commit 9068cdf
Show file tree
Hide file tree
Showing 4 changed files with 5 additions and 5 deletions.
Binary file modified docs/mutect/mutect.pdf
Binary file not shown.
4 changes: 2 additions & 2 deletions docs/mutect/mutect.tex
Original file line number Diff line number Diff line change
Expand Up @@ -133,7 +133,7 @@ \subsection{Finding Active Regions}

Under the first assumption the contribution of ref bases to the sum in Equation \ref{tumor-lod} vanishes -- errorless ref bases contribute no entropy -- so we only need to sum over alt bases. By the second assumption the likelihood of an alt read is related to its base quality (and associated error rate $\epsilon$) as $\ell_{r,{\rm alt}} = 1 - \epsilon_r$.

We compute Equation \ref{tumor-lod} for two possibilities; first, that an alt allele exists, second that only the ref allele exists. In the former case, Initializing $\bar{z}_{r,{\rm ref(alt)}} = 1$ for ref (alt) bases a single iteration for $q(\vf)$ gives $\vbeta = (N_{\rm alt} + 1, N_{\rm ref} + 1)$. For alt reads $r$ we then obtain from Equation \ref{f-tilde}
We compute Equation \ref{tumor-lod} for two possibilities; first, that an alt allele exists, second that only the ref allele exists. In the former case, Initializing $\bar{z}_{r,{\rm ref(alt)}} = 1$ for ref (alt) bases a single iteration for $q(\vf)$ gives $\vbeta = (N_{\rm alt} + 1, N_{\rm ref} + 1)$. For alt reads $r$ we then obtain from Equation \ref{z-bar}
\begin{equation}
\bar{z}_{r,{\rm alt}} = \frac{\tilde{f}_{\rm alt} (1 - \epsilon_r)}{\tilde{f}_{\rm alt} (1 - \epsilon_r) + \tilde{f}_{\rm ref} \epsilon_r},
\end{equation}
Expand Down Expand Up @@ -166,7 +166,7 @@ \subsection{Somatic Likelihoods Model}\label{introduction}
\end{align}
where, with $\psi$ denoting the digamma function, the moments
\begin{align}
\bar{z}_{ra} =& E_{q(\vz)} \left[ z_{ra} \right] = \frac{\tilde{f}_a \ell_{ra}}{\sum_{a^\prime} \tilde{f}_{a^\prime} \ell_{ra^\prime}} \\
\bar{z}_{ra} =& E_{q(\vz)} \left[ z_{ra} \right] = \frac{\tilde{f}_a \ell_{ra}}{\sum_{a^\prime} \tilde{f}_{a^\prime} \ell_{ra^\prime}} \label{z-bar} \\
\ln \tilde{f}_a =& E_{q(\vf)}[\ln f_a] = \psi(\beta_a) - \psi(\sum_{a^\prime} \beta_{a^\prime})
\label{f-tilde}
\end{align}
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
import org.apache.logging.log4j.Logger;

/**
* A helper class to maintain a cache of an int to double function
* A helper class to maintain a cache of an int to double function defined on n = 0, 1, 2. . .
* The cache expands when a number is not available.
* NOTE: this cache is thread safe and it may be accessed from multiple threads.
*/
Expand All @@ -20,7 +20,7 @@ public abstract class IntToDoubleFunctionCache {
public IntToDoubleFunctionCache() { }

/**
* Get the value of log10(n), expanding the cache as necessary
* Get the value of the function, expanding the cache as necessary. The cache only applies to non-negative values.

This comment has been minimized.

Copy link
@samuelklee

samuelklee Mar 25, 2019

Contributor

return needs updating too

This comment has been minimized.

Copy link
@davidbenjamin

davidbenjamin Mar 25, 2019

Author Contributor

done

* @param i operand
* @return log10(n)
*/
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@
/**
* Wrapper class so that the log10Factorial array is only calculated if it's used
*/
public class Log10FactorialCache extends IntToDoubleFunctionCache {
public final class Log10FactorialCache extends IntToDoubleFunctionCache {
private static final int CACHE_SIZE = 10_000;

@Override
Expand Down

0 comments on commit 9068cdf

Please sign in to comment.