diff --git a/NEWS.md b/NEWS.md
index fb3ab4d..b229200 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,3 +1,17 @@
+Release 0.6-r185 (12 December 2022)
+-----------------------------------
+
+Notable changes:
+
+ * Improvement: for each protein, only output alignments close to the best
+   alignment. Also added option --outs to tune the threshold.
+
+ * New feature: output GTF with option --gtf.
+
+(0.6: 22 December 2022, r185)
+
+
+
 Release 0.5-r179 (17 October 2022)
 ----------------------------------
 
diff --git a/miniprot.1 b/miniprot.1
index 8a203e2..27aede2 100644
--- a/miniprot.1
+++ b/miniprot.1
@@ -1,4 +1,4 @@
-.TH miniprot 1 "17 October 2022" "miniprot-0.5 (r179)" "Bioinformatics tools"
+.TH miniprot 1 "12 December 2022" "miniprot-0.6 (r185)" "Bioinformatics tools"
 .SH NAME
 .PP
 miniprot - protein-to-genome alignment with splicing and frameshifts
diff --git a/miniprot.h b/miniprot.h
index 5bb5438..4e298ca 100644
--- a/miniprot.h
+++ b/miniprot.h
@@ -3,7 +3,7 @@
 
 #include <stdint.h>
 
-#define MP_VERSION "0.5-r182-dirty"
+#define MP_VERSION "0.6-r185"
 
 #define MP_F_NO_SPLICE    0x1
 #define MP_F_NO_ALIGN     0x2
diff --git a/tex/miniprot.tex b/tex/miniprot.tex
index 26864b6..34d53b9 100644
--- a/tex/miniprot.tex
+++ b/tex/miniprot.tex
@@ -438,7 +438,7 @@ \subsection{Evaluated tools}
 
 To evaluate what aligners can map proteins to a whole genome, we randomly
 sampled 1\% of zebrafish proteins and mapped with various aligners. Only
-miniprot-0.5, Spaln2-2.4.13c~\citep{Iwata:2012aa}, GeMoMa-1.9~\citep{Keilwagen:2019wz}
+miniprot-0.6, Spaln2-2.4.13c~\citep{Iwata:2012aa}, GeMoMa-1.9~\citep{Keilwagen:2019wz}
 GenomeThreader-1.7.3~\citep{DBLP:journals/infsof/GremmeBSK05} could finish the
 alignment in an hour.  GenomeThreader found less than 30\% of coding regions in
 Spaln2 or miniprot alignment. It is not sensitive enough for the human-fish
@@ -538,10 +538,10 @@ \subsection{Evaluating protein-to-genome alignment}
 careful algorithm.
 
 Table~\ref{tab:eval} only considers the best hit of each protein. Miniprot by
-default may output multiple suboptimal alignments. If we count all
-human-zebrafish alignments, we could improve the base sensitivity to 65.32\%
-but with junction accuracy dropped to 90.87\%. The base specificity drops
-further to 84.96\% because miniprot starts to report pseudogenes.
+default may output multiple suboptimal alignments per protein if their
+alignment scores are no less than 99\% of the best alignment. If we count all
+human-zebrafish alignments outputted by minimap2, we could improve the base
+sensitivity to 60.76\% with a minor cost on base specificity to 95.25\%.
 
 \section{Discussions}