v1.4.1.5

malabz · Aug 22, 2023 · 355a188 · 355a188
1 parent 1cbade4
commit 355a188
Show file tree

Hide file tree

Showing 9 changed files with 70 additions and 10 deletions.
diff --git a/README.md b/README.md
@@ -13,6 +13,11 @@
 
 Modified by wym6912 wym6912@outlook.com
 
+## Updates (v1.4.1.5)
+
+- Try to fix a bug on processing 0 length sequence
+- Add sequence comment print
+
 ## Updates (v1.4.1.4)
 
 - Fix a critical bug on freeing SIMD varibles in Windows programs
@@ -58,8 +63,8 @@ abPOA is not only a stand-alone tool for MSA and consensus calling, it can also
 
 Download the [latest release](https://github.com/malabz/abPOA/releases):
 ```bash
-wget https://github.com/malabz/abPOA/releases/download/v1.4.1.3/abPOA-v1.4.1.3.tar.gz
-tar -zxvf abPOA-v1.4.1.3.tar.gz && cd abPOA-v1.4.1.3
+wget https://github.com/malabz/abPOA/releases/download/v1.4.1.5/abPOA-v1.4.1.5.tar.gz
+tar -zxvf abPOA-v1.4.1.5.tar.gz && cd abPOA-v1.4.1.5
 ```
 Make from source and run with test data:
 ```bash
@@ -73,7 +78,7 @@ abpoa ./test_data/seq.fa > cons.fa
 
 ### Windows
 
-Download the [latest release](https://github.com/malabz/abPOA/releases/download/v1.4.1.1/abPOA-v1.4.1.1.tar.gz) and use `Bandizip`, `WinRAR` or any other archive manager software.
+Download the [latest release](https://github.com/malabz/abPOA/releases/download/v1.4.1.5/abPOA-v1.4.1.5.tar.gz) and use `Bandizip`, `WinRAR` or any other archive manager software.
 
 Make programs from `Visual Studio 2022`:
 - First of all, download and install [`Visual Studio 2022`](https://visualstudio.microsoft.com/vs/).
@@ -111,8 +116,20 @@ abpoa ./test_data/seq.fa > cons.fa
 ## Table of Contents
 
 - [abPOA: adaptive banded Partial Order Alignment](#abpoa-adaptive-banded-partial-order-alignment)
-  - [Updates (v1.4.1)](#updates-v141)
+- [Modified by wym6912](#modified-by-wym6912)
+  - [Updates (v1.4.1.5)](#updates-v1415)
+  - [Updates (v1.4.1.4)](#updates-v1414)
+  - [Updates (v1.4.1.3)](#updates-v1413)
+  - [Updates (v1.4.1.2)](#updates-v1412)
+  - [Updates (v1.4.1.1)](#updates-v1411)
+  - [Windows support](#windows-support)
+    - [For development in Windows](#for-development-in-windows)
   - [Getting started](#getting-started)
+    - [Linux](#linux)
+    - [Windows](#windows)
+- [RAW README](#raw-readme)
+  - [Updates (v1.4.1)](#updates-v141)
+  - [Getting started](#getting-started-1)
   - [Table of Contents](#table-of-contents)
   - [Introduction](#introduction)
   - [Installation](#installation)
@@ -140,7 +157,6 @@ abpoa ./test_data/seq.fa > cons.fa
   - [For development](#for-development)
   - [Evaluation datasets](#evaluation-datasets)
   - [Contact](#contact)
-  - [Windows modification](#windows-modification)
 
 ## <a name="introduction"></a>Introduction
 abPOA is an extended version of [Partial Order Alignment (POA](10.1093/bioinformatics/18.3.452)) 

diff --git a/abPOA.rc b/abPOA.rc
diff --git a/abPOA_profile.rc b/abPOA_profile.rc
diff --git a/src/abpoa.c b/src/abpoa.c
@@ -20,7 +20,7 @@ char PROG[20] = "abpoa";
 #define _bO BOLD UNDERLINE "O" NONE
 #define _bA BOLD UNDERLINE "A" NONE
 char DESCRIPTION[200] = _ba "daptive " _bb "anded " _bP "artial " _bO "rder " _bA "lignment";
-char VERSION[20] = "1.4.1.4";
+char VERSION[20] = "1.4.1.5";
 char CONTACT[30] = "gaoy1@chop.edu";
 
 const struct option abpoa_long_opt [] = {

diff --git a/src/abpoa_align.c b/src/abpoa_align.c
@@ -364,6 +364,7 @@ int abpoa_poa(abpoa_t *ab, abpoa_para_t *abpt, uint8_t **seqs, int **weights, in
     uint8_t *qseq, *rc_qseq; int *weight, *rc_weight;
     // uint8_t *seq1;
     for (i = 0; i < n_seq; ++i) {
+        if(seq_lens[i] == 0) continue;
         qlen = seq_lens[i]; qseq = seqs[i]; weight = weights[i]; read_id = exist_n_seq + i;
 #ifdef __DEBUG__
         fprintf(stderr, "seq: # %d\n", i);
@@ -507,9 +508,39 @@ int abpoa_msa1(abpoa_t *ab, abpoa_para_t *abpt, char *read_fn, FILE *out_fp) {
     // always reset graph before perform POA
     int max_len = 0;
     for (i = 0; i < abs->n_seq; ++i) {
+        if (! abs->seq[i].l)
+        {
+            _err_simple_func_printf("Warning: Found a sequence length = 0. Program maybe generate wrong consensus sequence and wrong gfa.");
+            continue;
+        }
+        else if (abs->seq[i].l == 1 && abs->seq[i].s[0] == 13)
+        {
+            _err_simple_func_printf("Warning: Found a sequence length = 0. Program maybe generate wrong consensus sequence and wrong gfa.");
+            continue;
+        }
         if (abs->seq[i].l > max_len) max_len = abs->seq[i].l;
     }
 
+    if (max_len == 0)
+    {
+        if (abpt->out_cons)
+            err_printf("Warning: no consensus sequence generated.\n");
+        if (abpt->out_gfa)
+            err_printf("Warning: no gfa generated.\n");
+        if (abpt->out_msa)
+        {
+            for(i = 0; i < abs->n_seq; ++i)
+            {
+                if (abs->name[i].l + abs->comment[i].l > 0) {
+                    fprintf(out_fp, ">%s %s\n", abs->name[i].l ? abs->name[i].s : "", abs->comment[i].l ? abs->comment[i].s : "");
+                } else {
+                    fprintf(out_fp, ">Seq_%d\n", i+1);
+                }
+                fputc('\n', out_fp);
+            }
+        }
+        return 0;
+    }
     // detect u and t
     int has_st = 0, has_su = 0;
 

diff --git a/src/abpoa_output.c b/src/abpoa_output.c
@@ -77,9 +77,9 @@ void abpoa_output_rc_msa(abpoa_t *ab, abpoa_para_t *abpt, FILE *out_fp) {
     abpoa_seq_t *abs = ab->abs; abpoa_cons_t *abc = ab->abc;
     if (abc->msa_len <= 0) return;
     for (i = 0; i < abs->n_seq; ++i) {
-        if (abs->name[i].l > 0) {
-            if (abs->is_rc[i]) fprintf(out_fp, ">%s_reverse_complement\n", abs->name[i].s);
-            else fprintf(out_fp, ">%s\n", abs->name[i].s);
+        if (abs->name[i].l + abs->comment[i].l > 0) {
+            if (abs->is_rc[i]) fprintf(out_fp, ">%s_reverse_complement %s\n", abs->name[i].l ? abs->name[i].s : "", abs->comment[i].l ? abs->comment[i].s : "");
+            else fprintf(out_fp, ">%s %s\n", abs->name[i].l ? abs->name[i].s : "", abs->comment[i].l ? abs->comment[i].s : "");
         } else {
             fprintf(out_fp, ">Seq_%d\n", i+1);
         }

diff --git a/src/abpoa_profile.c b/src/abpoa_profile.c
@@ -21,7 +21,7 @@ char PROG[20] = "abpoa_profile";
 #define _bO BOLD UNDERLINE "O" NONE
 #define _bA BOLD UNDERLINE "A" NONE
 char DESCRIPTION[200] = _ba "daptive " _bb "anded " _bP "artial " _bO "rder " _bA "lignment profile-profile mode";
-char VERSION[20] = "1.4.1.4";
+char VERSION[20] = "1.4.1.5";
 char CONTACT[30] = "gaoy1@chop.edu";
 
 int need_prealign;

diff --git a/test_data/all_0_len.fasta b/test_data/all_0_len.fasta
@@ -0,0 +1,5 @@
+> seq1
+
+> seq2
+
+> seq3
diff --git a/test_data/has_0_len.fasta b/test_data/has_0_len.fasta
@@ -0,0 +1,8 @@
+> seq1
+
+> seq2
+AC
+> seq3
+
+> seq4
+AG
-Original file line number
+Diff line change
@@ -0,0 +1,5 @@
+    > seq1
+    > seq2
+    > seq3